Daniel S. Weld
Daniel Sabey Weld is an American computer scientist who is the Thomas J. Cable/WRF Professor of Computer Science and Engineering at the University of Washington, where he does research in automated planning and scheduling, software agents, and Internet information extraction. He is a venture partner at Madrona Venture Group, a Seattle-based venture capital firm. More on Wikipedia
-
Daniel Weld reposted
Chenhao Tan
@ChenhaoTan
-
Daniel Weld
@dsweld
This benchmark for Ai scientific capabilities is beautifully thought out - I especially like it's clear enumeration of design principles...
-
Daniel Weld reposted
Zixian Ma
@zixianma02
We built MolmoWeb from the scratch with Molmo2!!! 💕🌐
It’s not easy to build SOTA web agents out of open source VLMs, when they can be so profitable that very few projects release everything (if anything), esp the datasets 🔑
But, we just released all the MolmoWeb model checkpoints and datasets from ai2😉
Can’t wait to see what the community builds on top of MolmoWeb!🫡 -
Daniel Weld reposted
Ai2
@allen_ai
🔎 Deep research agents like Asta ScholarQA and OpenAI Deep Research are transforming how we perform literature review.
But how do we know if the way we evaluate them is actually meaningful?
Announcing our new paper: “Deep Research, Shallow Evaluation: A Case Study in Meta-Evaluation for Long-Form QA Benchmarks” 🧵 -
Daniel Weld reposted
Pao Siangliulue
@Siangliulue
Are you a researcher in CS or a CS-adjacent field curious about how an AI agent can help you with your research project? Want to try a new tool for your research support in a paid user study ($100, 2 hr)? Limited spot numbers. See details and sign up here: forms.gle/JzLtkAhe7TtvuiwQ8
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Truly open scientific question answering - that's good!
www.science.org/content/article/open-source-ai-program-ca… -
Daniel Weld reposted
Ai2
@allen_ai
We’re releasing the Theorizer code and framework + a dataset of ~3,000 theories generated by Theorizer across the field of AI/NLP, built from 13,744 source papers.
💻 Code: github.com/allenai/asta-theorizer
📝 Technical report: arxiv.org/abs/2601.16282
✍️ Learn more in our blog: -
Daniel Weld
@dsweld
I'm so excited by this! Our system is generating some insightful & novel theories (e.g., internally for LM post-training). And it's still getting better!
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Smart analysis analysis of scholar output when authors adopted LLMs as part of their writing: 1) huge 36% boost in # papers published 2) LLMs mitigate skill disparities, eg native language - enough to shift market share of production toward China www.semanticscholar.org/paper/Scientific-production-in-th… @yian_yin
-
Daniel Weld reposted
-
Daniel Weld reposted
Ai2
@allen_ai
🧠 Introducing NeuroDiscoveryBench. Built with @AllenInstitute, it’s the first benchmark for evaluating AI systems like our Asta DataVoyager agent on neuroscience data. The benchmark tests whether AI can truly extract insights from complex brain datasets.
-
Daniel Weld reposted
Bodhisattwa Majumder
@mbodhisattwa
#neurips2025 and AI x Science?
Some fun announcements are coming up. Stay tuned.
Also, our Asta internship application is still open -- apply and mention my name if you'd like to work w me ~
Bodhisattwa Majumder
@mbodhisattwa
🎀 Really excited to start cadence for 2026!
At @allen_ai, we are at the intersection of AI making a real impact in accelerating science, with several serious collaborations with domain scientists (Economics, Oncology, Neuroscience, Climate, Epidemiology).
If you are passionate about turning the stack upside down to break and build current LLMs to be adaptive for a continual discovery process, join us!
If you are interested in my work, such as data-driven discovery, open-ended hypothesis search, test-time adoption for hypothesis generation, causal mechanism discovery using data and literature, mention my name in your application!
Past interns might vouch for their experiences, but working with Asta interns for the last two years has been one of my most rewarding journeys at Ai2! 🩷 -
Daniel Weld reposted
-
Daniel Weld reposted
Bodhisattwa Majumder
@mbodhisattwa
Plenty of AI-gen papers in ICLR. Wonder why?
🚨 In a preregistered Randomized Controlled Trial, we find: CS authors perceive AI-abstracts as more readable, tend to edit less than their published counterparts. AI-use and its disclosure shape the fabric of collaborative scientific writing.
Work led by @hsanchaita & @leadoeun27, advised by @shocheen & yours truly.
1/n -
Daniel Weld
@dsweld
Impressive deep-research performance by a tiny & open model!
Rulin Shao
@RulinShao
🔥Thrilled to introduce DR Tulu-8B, an open long-form Deep Research model that matches OpenAI DR 💪Yes, just 8B! 🚀
The secret? We present Reinforcement Learning with Evolving Rubrics (RLER) for long-form non-verifiable DR tasks! Our rubrics:
- co-evolve with the policy model
- are grounded on search knowledge
🧵 -
Daniel Weld
@dsweld
The benchmark desiderata alone make this paper worth a read...
Jonathan Bragg
@turingmusician
Agent benchmarks don't measure true *AI* advances
We built one that's hard & trustworthy
👉AstaBench tests agents w/ *standardized tools* on 2400+ scientific research problems
👉SOTA results across 22 agent *classes*
👉AgentBaselines agents suite
🆕arxiv.org/abs/2510.21652
🧵👇 -
Daniel Weld
@dsweld
Super interesting and well written summary of the incredible progress we’ve made on climate change (and what’s most important to do next) ⭐️⭐️⭐️⭐️⭐️ www.gatesnotes.com/three-tough-truths-about-climate
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Pretty amazing that this can be done at all, but especially with federated data (crucial given the sensitivity of patient data)!
-
Daniel Weld
@dsweld
Very excited to share: www.geekwire.com/2025/an-open-source-ai-sidekick-for-scie…
-
Daniel Weld reposted
Weijia Shi
@WeijiaShi2
Can data owners & LM developers collaborate to build a strong shared model while each retaining data control?
Introducing FlexOlmo💪, a mixture-of-experts LM enabling:
• Flexible training on your local data without sharing it
• Flexible inference to opt in/out your data anytime
At 37B parameters, FlexOlmo is competitive across 31 tasks. -
Daniel Weld reposted
John Nguyen
@__JohnNguyen__
AI2 Paper Finder and AI2 ScholarQA are the best tools for finding related works. I'm surprised they aren't more popular @allen_ai
-
Daniel Weld reposted
Aaron Tay
@aarontay
The trouble with finding a specific papers you read but forgetten is there's a diff between 1)what the paper says 2) what other people cite it for 3) what YOU want to cite it for! Trying ai2 paper finder makes it clear ha
Aaron Tay
@aarontay
This might be useful to find papers on specific things but you don't remember the exact reference to
-
Daniel Weld
@dsweld
Need someone (or something!) to amplify the speed of your ML/NLP experimentation rate?
-
Daniel Weld
@dsweld
Use the power of LLMs to find academic papers for your hardest queries!
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Check it out!
allenai.org/blog/ai2-scholarqa -
Daniel Weld
@dsweld
Try our new scientific question answering system!
Ai2
@allen_ai
Can AI really help with literature reviews? 🧐
Meet Ai2 ScholarQA, an experimental solution that allows you to ask questions that require multiple scientific papers to answer. It gives more in-depth, detailed, and contextual answers with table comparisons, expandable sections for subtopics, and citations with paper excerpts for verification 💡
Try Ai2 ScholarQA now: scholarqa.allen.ai
More in 🧵 -
Daniel Weld
@dsweld
This is still experimental, but ... really cool! Try it!
Joseph Chee Chang
@josephcc
Joint work with @aps6992 @arnaik19 @SergeyFeldman @Siangliulue @kylelostat @_DougDowney @dsweld and more. Please check it out and give us feedback : )
Ai2
@allen_ai
Can AI really help with literature reviews? 🧐
Meet Ai2 ScholarQA, an experimental solution that allows you to ask questions that require multiple scientific papers to answer. It gives more in-depth, detailed, and contextual answers with table comparisons, expandable sections for subtopics, and citations with paper excerpts for verification 💡
Try Ai2 ScholarQA now: scholarqa.allen.ai
More in 🧵 -
Daniel Weld reposted
Joseph Chee Chang
@josephcc
Joint work with @aps6992 @arnaik19 @SergeyFeldman @Siangliulue @kylelostat @_DougDowney @dsweld and more. Please check it out and give us feedback : )
Ai2
@allen_ai
Can AI really help with literature reviews? 🧐
Meet Ai2 ScholarQA, an experimental solution that allows you to ask questions that require multiple scientific papers to answer. It gives more in-depth, detailed, and contextual answers with table comparisons, expandable sections for subtopics, and citations with paper excerpts for verification 💡
Try Ai2 ScholarQA now: scholarqa.allen.ai
More in 🧵 -
Daniel Weld
@dsweld
More great work by FutureHouse!
Andrew White 🐦⬛
@andrewwhite01
Finishing 2024 with one more research result! We’ve trained small language agents to do hard sci tasks: engineering proteins, manipulating DNA, and working with sci literature in a new library called Aviary. We beat humans and frontier LLMs on these tasks!
-
Daniel Weld
@dsweld
Exciting chance to work with a phenomenal team, producing open AI tech for the common good! Apply now!
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar
is hiring #nlproc #hci #ml #ai researchers for a Research Engineer position with target start dates in 2025, apply by *Jan 17, 2025*!
Apply: job-boards.greenhouse.io/thealleninstitute/jobs/6481983 -
Daniel Weld
@dsweld
Nice work on a method and UI for human-ai teaming!
arxiv.org/abs/2412.10999
Task is the process of performing scientific research, but technique is general. With @kjfeng_ @josephcc @amyxzh and many more -
Daniel Weld
@dsweld
Curious people's reaction to this... How should one design for effective human-AI communication?
Gagan Bansal
@bansalg_
☀️Excited to share our new paper!
Generative AI agents are powerful but complex—how do we design them for transparency and human control? 🤖✨
At the heart of this challenge is establishing common ground, a concept from human communication. Our new paper identifies 12 key challenges in improving common ground between humans and AI agents.
1/ -
Daniel Weld
@dsweld
Great review of OpenScholar venturebeat.com/ai/openscholar-the-open-source-a-i-thats-…
Try it at: openscholar.allen.ai -
Daniel Weld reposted
Akari Asai
@AkariAsai
3/ 🔍 What is OpenScholar?
It's a retrieval-augmented LM with
1️⃣ a datastore of 45M+ open-access papers
2️⃣ a specialized retriever and reranker to search the datastore
3️⃣ an 8B Llama fine-tuned LM trained on high-quality synthetic data
4️⃣ a self-feedback generation pipeline -
Daniel Weld
@dsweld
Very cool scientific QA system - try it!
Akari Asai
@AkariAsai
1/ Introducing ᴏᴘᴇɴꜱᴄʜᴏʟᴀʀ: a retrieval-augmented LM to help scientists synthesize knowledge 📚 @uwnlp @allen_ai
With open models & 45M-paper datastores, it outperforms proprietary systems & match human experts.
Try out our demo!
We also introduce ꜱᴄʜᴏʟᴀʀQᴀʙᴇɴᴄʜ, a new large-scale multi-domain benchmark for scientific research synthesis, covering CS, Bio and Physics.This media couldn't be displayed. See it on the original post. -
Daniel Weld
@dsweld
Fantastic opportunity to do cutting edge work on LLM agents - in an open, nonprofit environment!
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar
is hiring #nlproc #hci #ml #ai researchers for an Applied Research Scientist position with target start dates in 2025, apply by *Nov 15*!
Apply: job-boards.greenhouse.io/thealleninstitute/jobs/6367281 -
Daniel Weld reposted
President Biden Archived
@POTUS46Archive
When I took office, we faced the worst economic crisis since the Great Depression.
Today, we have the greatest economy in the world. -
Daniel Weld reposted
Erik Brynjolfsson
@erikbryn
The American Economy has become the envy of the world
🚀🚀🚀🇺🇸🇺🇸🇺🇸 -
Daniel Weld reposted
Bodhisattwa Majumder
@mbodhisattwa
Makes sense! We push with full force in this direction:
Bodhisattwa Majumder
@mbodhisattwa
Can LLMs help accelerate the discovery of data-driven scientific hypotheses? 🧬📊
We benchmark this in DiscoveryBench: 264 discovery tasks from 6 scientific domains, from humanities to biology: arxiv.org/pdf/2407.01725v1
@allen_ai @ai2_aristo
@surana_h @UMassAmherst -
Daniel Weld
@dsweld
Come join us at Ai2! Tackle challenging problems with great mentors & phenomenal resources - for the common good with a truly open philosophy and emphasis on publishing!
Semantic Scholar Research @ AI2
@ai2_s2research
Reminder: the 1st rolling deadline to apply for our open positions is this Friday (Nov 1)!
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar is hiring #nlproc #hci #ml #ai researchers for the following positions with target start dates in 2025, apply by *Nov 1* for the 1st rolling deadline.
- Research intern
- Young investigator (Postdoc)
- Research scientist
Apply: job-boards.greenhouse.io/thealleninstitute?departments[]=… -
Daniel Weld reposted
Ben Lee
@lee_bcg
I’m recruiting Ph.D. students to join the Lab for Computing Cultural Heritage @uw_ischool! If you’re interested in working with me and pursuing a Ph.D. at the intersection of AI, LIS, and the digital humanities, you can find more information here: bcglee.com/lcch.html
-
Daniel Weld reposted
Ai2
@allen_ai
We're hiring OLMo research interns 📢
Apply by 11/3 for Spring/Summer 2025 internships!
Nathan Lambert
@natolambert
We're hiring research interns at Ai2 to help at every stage of our Open Language Model (OLMo) workstream. We want people who can contribute meaningfully to our best models.
The best thing you can do is apply through the link below.
The bar is very high, I cannot reply to everyone who reaches out.
If you want to reach out directly to me (or another team member) about this:
1. I recommend email.
2. Be brief.
3. Make it clear why you have the ability or experience to be a meaningful contributor. No need for fluff.
I'll be mentoring directly the right person on post-training, but involved generally with most of the workstreams to some level. Really excited to build truly open AI with some new friends 🚀 -
Daniel Weld reposted
Joseph Chee Chang
@josephcc
Hello #uist2024 ! Ai2 / Semantic Scholar Research is looking for summer interns again :) If you’re interested in working with me next summer come chat about interactive search and synthesis, literature review support tools, and LLM agents : )
Kyle Lo
@kylelostat
If you're at #uist2024 and interested in augmented reading and/or related topics in using AI to provide research assistance, @SemanticScholar @allen_ai is hiring 🤩
@josephcc is there in person, go find him for a chat! -
Daniel Weld reposted
Kyle Lo
@kylelostat
excited that our Semantic Reader paper is published in Communications of the ACM! 🥳
📰cacm.acm.org/research/the-semantic-reader-project
📽️youtu.be/W-6qouywnIk
Kyle Lo
@kylelostat
The Semantic Reader project combines AI & HCI research to explore the future of scientific reading. We ask: “Can we create intelligent, interactive, and accessible reading interfaces for research papers, even atop existing PDFs?” www.semanticscholar.org/reader/67a5bacb00651dbe0dd9ef2a56… 1/n
-
Daniel Weld reposted
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar is hiring #nlproc #hci #ml #ai researchers for the following positions with target start dates in 2025, apply by *Nov 1* for the 1st rolling deadline.
- Research intern
- Young investigator (Postdoc)
- Research scientist
Apply: job-boards.greenhouse.io/thealleninstitute?departments[]=… -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
📢 Are you a #nlproc #hci #ml #ai researcher excited about accelerating science with AI?
We're #hiring for Research Scientist, Young Investigator (Postdoc Researcher), and Research Intern with start dates in 2025!
job-boards.greenhouse.io/thealleninstitute?departments[]=…
Apply by 11/1 for full consideration ⏰ -
Daniel Weld reposted
Kyle Lo
@kylelostat
🥳 yay it's finally out -- olmo with eyes! 👀
🐙molmo is our family of open, late-fusion image+text language models, all available on huggingface 🤗
🐟trained using a really high-quality dataset of images + dense captions / task demonstrations that we'll be releasing openly
Ai2
@allen_ai
Meet Molmo: a family of open, state-of-the-art multimodal AI models.
Our best model outperforms proprietary systems, using 1000x less data.
Molmo doesn't just understand multimodal data—it acts on it, enabling rich interactions in both the physical and virtual worlds.
Try it for yourself: molmo.allenai.orgThis media couldn't be displayed. See it on the original post. -
Daniel Weld
@dsweld
Q: Can AI *really* help scientists develop better research ideas? ... A: Yes!
Marissa Radensky
@marissa_rad
Can recombining aspects of analogous papers and your own interests help you find your next research project? Introducing Scideator 🧪💡, a mixed-initiative tool for scientific ideation that works with you to generate ideas 🔀 and evaluate/improve their novelty 🔍.
-
Daniel Weld
@dsweld
Looks really exciting!
Sam Rodriques
@SGRodriques
Also, see our preprint for details, here: paper.wikicrow.ai 5/
-
Daniel Weld reposted
Marissa Radensky
@marissa_rad
We're seeking CS researchers to participate in a study on working with AI tools to come up with research project ideas!
Recruitment survey: forms.gle/A7G5ZFpGYC5Re86r8
Compensation: $50 Amazon gift card
Time: 85 mins (includes 5-min break)
@dsweld @Hoper_Tom @Siangliulue @rayrayfok -
Daniel Weld reposted
Arthur Douillard
@Ar_Douillard
"White House says no need to restrict ‘open-source’ artificial intelligence — at least for now"
apnews.com/article/ai-open-source-white-house-f62009172c4… -
Daniel Weld reposted
Xingyao Wang
@xingyaow_
Software is a powerful tool, enabling human developers to interact with the world in complex & profound ways. What if we could use software as a tool to create similar versatile AI agents? Meet OpenDevin: an open platform for AI software developers as generalist agents.
🧵 1/ -
Daniel Weld
@dsweld
An incredible resource for the important task of scientific discovery!
Bodhisattwa Majumder
@mbodhisattwa
Can LLMs help accelerate the discovery of data-driven scientific hypotheses? 🧬📊
We benchmark this in DiscoveryBench: 264 discovery tasks from 6 scientific domains, from humanities to biology: arxiv.org/pdf/2407.01725v1
@allen_ai @ai2_aristo
@surana_h @UMassAmherst -
Daniel Weld reposted
Steven Pinker
@sapinker
One of @EasterbrookG's collection of headlines showing media negativity bias (e.g., Economy Improving, But Some Left Behind; Not Everyone in World's Happiest Country are Happy).
Gregg Easterbrook
@EasterbrookG
Financial Times headline: WORLD FACES STAGGERING OIL GLUT, ENERGY WATCHDOG WARNS.
Warns! Warns a resource is plentiful! P.S. if were an oil shortage, MSM would say that’s bad too. -
Daniel Weld reposted
Surya Ganguli
@SuryaGanguli
“Imagine if we passed a law that says if anyone uses a motor in a harmful way, the *motor manufacturer* is liable.” This would be ridiculous. Well that is what California’s SB-1047 is proposing for AI models. Regulate the application. Not the math.
Andrew Ng
@AndrewYNg
The effort to protect innovation and open source continues. I believe we’re all better off if anyone can carry out basic AI research and share their innovations. Right now, I’m deeply concerned about California's proposed law SB-1047. It’s a long, complex bill with many parts that require safety assessments, shutdown capability for models, and so on.
There are many things wrong with this bill, but I’d like to focus here on just one: It defines an unreasonable “hazardous capability” designation that may make builders of large AI models liable if someone uses their models to do something that exceeds the bill’s definition of harm (such as causing $500 million in damage). That is practically impossible for any AI builder to ensure. If the bill is passed in its present form, it will stifle AI model builders, especially open source developers.
Some AI applications, for example in healthcare, are risky. But as I wrote previously, regulators should regulate applications rather than technology.
- Technology refers to tools that can be applied in many ways to solve various problems.
- Applications are specific implementations of technologies designed to meet particular customer needs.
For example, an electric motor is a technology. When we put it in a blender, an electric vehicle, dialysis machine, or guided bomb, it becomes an application. Imagine if we passed laws saying, if anyone uses a motor in a harmful way, the motor manufacturer is liable. Motor makers would either shut down or make motors so tiny as to be useless for most applications. If we pass such a law, sure, we might stop people from building guided bombs, but we’d also lose blenders, electric vehicles, and dialysis machines. In contrast, if we look at specific applications, like blenders, we can more rationally assess risks and figure out how to make sure they’re safe, and even ban classes of applications, like certain types of munitions.
Safety is a property of the application, not a property of the technology (or model), as @random_walker and @sayashk have pointed out. Whether a blender is a safe one can’t be determined by examining the electric motor. A similar argument holds for AI.
SB-1047 doesn’t account for this distinction. It ignores the reality that the number of beneficial uses of AI models is, like electric motors, vastly greater than the number of harmful ones. But, just as no one knows how to build a motor that can’t be used to cause harm, no one has figured out how to make sure an AI model can’t be adapted to harmful uses. In the case of open source models, there’s no known defense to fine-tuning to remove RLHF alignment. And jailbreaking work has shown that even closed-source, proprietary models that have been properly aligned can be attacked in ways that make them give harmful responses. Indeed, the sharp-witted @elder_plinius regularly tweets about jailbreaks for closed models. Kudos also to Anthropic’s @cem__anil and collaborators for publishing their work on many-shot jailbreaking, an attack that can get leading large language models to give inappropriate responses and is hard to defend against.
California has been home to a lot of innovation in AI. I’m worried that this anti-competitive, anti-innovation proposal has gotten so much traction in the legislature. Worse, other jurisdictions often follow California, and it would be awful if they were to do so in this instance.
SB-1047 passed in a key vote in the State Senate in May, but it still has additional steps before it becomes law. I hope you will speak out against it if you get a chance to do so.
[Original text (with links): www.deeplearning.ai/the-batch/issue-252/ ] -
Daniel Weld reposted
Startup Archive
@StartupArchive_
NVIDIA CEO Jensen Huang: “I really discourage 1-on-1s”
Jensen famously has 60 direct reports. When Stripe founder Patrick Collison points out that this isn’t conventionally considered best practice, Jensen shares his reasoning:
“I don’t do 1-on-1s, and almost everything I say, I say to everybody all the time. I don’t really believe there’s any information that I operate on that only one or two people should hear about… I believe that when you give everybody equal access to information, that empowers people. And so that’s number one… Number two, if the CEO’s direct staff is 60 people, the number of layers you’ve removed in a company is probably something like seven.”
Patrick offers to steal man the other side of the argument:
“1-on-1s are where you provide coaching, where you maybe talk through personal goals and career advancement, where maybe you give feedback on something that you see somebody systematically not doing so well… Do you not do those things or do you do them in a different way?”
Jensen responds:
“I give you feedback right there in front of everybody. In fact, this is a really big deal. First of all, feedback is learning. For what reason are you the only person who should learn this?… We should all learn from that opportunity… Half the time I’m not right, but for me to reason through it in front of everybody helps everybody learn how to reason through it. The problem I have with 1-on-1s and taking feedback aside is you deprive a whole bunch of people that same learning. Learning from other people’s mistakes is the best way to learn.”
Video Source: @stripeThis media couldn't be displayed. See it on the original post. -
Daniel Weld
@dsweld
AI-powered scientific discovery?
Yes we can! blog.allenai.org/data-driven-discovery-with-large-generat… @allen_ai -
Daniel Weld reposted
Katie Keith
@katakeith
Excited for two upcoming opportunities where we’ll talk about our TMLR paper (which was published this fall) “RCT Rejection Sampling for Causal Estimation Evaluation” with coauthors @SergeyFeldman @david__jurgens @turingmusician @r0ntu openreview.net/pdf?id=F74ZZk5hPa
-
Daniel Weld reposted
Tal August
@tal_august
Excited to be presenting this work at #chi2024. If you're interested in how language technologies can help us adapt communication to different people, stop by bright and early at 9 am on Wednesday!
Tal August
@tal_august
1/n Language models (LMs) can support science communication by simplifying complex text for envisioned audiences ("write for 5th grader"). But simplifying can convey less information or lead to overconfidence. Our #chi2024 paper investigates when and who we should simplify for
-
Daniel Weld reposted
Andrew Head
@drewmikehead
@PennHCI is coming to #chi2024! With not one but two talks about augmented reading 📖✨ Look out for these faces — Hita, @litao_yan, and I will be there and excited to catch up with you all :-)
-
Daniel Weld
@dsweld
Exciting insights about LLM chaining from work on crowdsourcing workflows! (Much improved from the last version of the paper!)
Madeleine Grunde-McLaughlin
@MadeleineGrunde
Chaining LLM calls can improve output quality, but navigating the massive space of task decompositions is challenging. Revisiting the established field of crowdsourcing, we distill strategies for effective LLM chain design and identify opportunities for future research. [1/11]
-
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
Technical academia (at least my little corner of it) thinks industry is wasting billions on a technology that is not fit for purpose. We are all using the open weights models and cheering for @allen_ai's fully open LLM effort.
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Looking to get more out of Semantic Scholar API? 🤩 Join our next webinar on 5/23 to learn about advanced uses, including different access modes, various use cases, self-hosting, and more!
Sign up now: forms.gle/7Q9pEBrrBhDBtDn56 -
Daniel Weld reposted
Sasha Rush
@srush_nlp
Talk: "OLMo: Findings of Training an Open LM" from Hanna Hajirshizi at AI2 from OSGAI.
Extremely interesting overview of the 4 parts (Data, Training, Adaptation, Eval) of the OLMo open LLM project. Rare insight into how these processes work at scale.
www.youtube.com/watch?v=qFZbu2P1vZ8 -
Daniel Weld
@dsweld
Impressed by @etzioni and @truemediadotorg's improtant work combating deepfakes! www.nytimes.com/2024/04/02/technology/an-ai-researcher-ta…
-
Daniel Weld reposted
Karen Hao
@_KarenHao
For years I’ve been interviewing data annotation workers who are the lifeblood of the AI industry. For years I’ve heard the same story: the platforms they work for wield total power, leaving them precarious & vulnerable to exploitation. A horrible example of this just happened 1/
-
Daniel Weld reposted
Kyle Lo
@kylelostat
can LMs help us write expository answers to scientific research questions?
excited to share our work led by @brunchavecmoi. we recruited NLP folks to work with an LM to answer research questions and logged successes/failures in sustained interaction traces🦉
Fangyuan Xu
@brunchavecmoi
Instruction-following capabilities of LLMs are a prerequisite to AI ✒️ writing assistance. How are good current LLMs at this task?
We present 🥝 𝗞𝗜𝗪𝗜, a dataset with instructions for knowledge-intensive, document-grounded writing for long-form answers to research questions. -
Daniel Weld reposted
Kyle Lo
@kylelostat
for those looking for instruction-tuned OLMo👇
Ai2
@allen_ai
Using our Open Instruct and Tulu 2, we adapt OLMo to acquire different capabilities and safety measures through fine-tuning and Direct Preference Optimization (DPO). The adapted models demonstrate quick improvement on popular reasoning tasks such as MMLU and TruthfulQA, and on safety datasets such as ToxicGen. huggingface.co/allenai/OLMo-7B-Instruct
-
Daniel Weld reposted
Arvind Narayanan
@random_walker
There is much speculation about the risks of releasing powerful AI models openly, but not enough rigorous analysis. To raise the quality of the debate, we brought together 25 authors from 16 academic, industry, and civil society organizations. Out today: crfm.stanford.edu/open-fms/
Sayash Kapoor
@sayashk
A recent MIT study claimed open models can help create bioweapons. But it didn’t test if they’re more useful than just having internet access (and later studies found they aren’t).
How can we assess the impact of open foundation models? New paper: crfm.stanford.edu/open-fms -
Daniel Weld reposted
Bodhisattwa Majumder
@mbodhisattwa
Is it possible to build end-to-end autonomous discovery systems using Large Generative Models (LGMs)? 🧬
In this position paper, we argue:
arxiv.org/pdf/2402.13610.pdf
🧵 (1/n)
@allen_ai @ai2_aristo
@surana_h @UMassAmherst @UUtah -
Daniel Weld reposted
Semantic Scholar Research @ AI2
@ai2_s2research
📣 Job opportunities at Semantic Scholar Research @ the Allen Institute for AI (AI2) for post-doctoral & pre-doctoral researchers starting in 2024! 📣
Our team works on NLP and HCI research with a focus on open LLMs and LLM-powered research support tools and assistants. -
Daniel Weld reposted
Iz Beltagy
@i_beltagy
OLMo-7b is finally out 🎉, and we are releasing everything; weights, intermediate checkpoints, training code and logs, training data and toolkit, evaluation and adaptation code and data.
Most of it has been released, and the rest is coming soon. OLMo-65b and Adapted OLMo-7b are training, and we plan extensions to future modalities. Check the report for details allenai.org/olmo/olmo-paper.pdf
Thanks to all of @allen_ai, especially those who took the project from zero-to-one: @mechanicaldirk, Pete Walsh, @akshitab93, Rodney Kinney, @oyvindtafjord, @AnanyaHarsh, @hamishivi, @IanMagnusson, @yizhongwyz, Kyle Richardson, @LukeZettlemoyer, @soldni, @nlpnoah, @HannaHajishirzi -
Daniel Weld reposted
Ai2
@allen_ai
OLMo is here! And it’s 100% open.
It’s a state-of-the-art LLM and we are releasing it with all pre-training data and code. Let’s get to work on understanding the science behind LLMs. Learn more about the framework and how to access it here:
blog.allenai.org/olmo-open-language-model-87ccfc95f580 -
Daniel Weld
@dsweld
Very exciting release of a small yet incredibly powerful LLM, with unparalleled openness, including all training data and details! techcrunch.com/2024/02/01/ai2-open-sources-text-generatin…
-
Daniel Weld
@dsweld
Important (and difficult) effort! Bravo!
toddbishop
@toddbishop
New nonpartisan AI nonprofit @truemediadotorg, led by Oren @Etzioni and backed by Uber co-founder Garrett Camp (@gc), is making a political deepfake detector www.geekwire.com/2024/new-ai-nonprofit-truemedia-led-by-o… via @GeekWire
-
Daniel Weld
@dsweld
I was joking about a free Uber service, but Bellevue's gone and done it
www.seattletimes.com/seattle-news/i-was-joking-about-a-fr… -
Daniel Weld reposted
Shaurya Rohatgi
@shauryr
Happy to see Semantic Scholar loader has more than 1k download on llamahub.ai 🙂 just above arxiv loader
-
Daniel Weld reposted
Ai2
@allen_ai
Today we joined leading AI partners to launch the pilot of @NSF’s National AI Research Resource (NAIRR), a shared research infrastructure that will democratize access to critical resources necessary to power responsible AI innovation. Read more here: blog.allenai.org/ai2-listed-as-founding-partner-of-nation…
U.S. National Science Foundation
@NSF
NSF and its partners are proud to launch the National AI Research Resource pilot.
Its goal? To democratize the future of #ai research & development by offering researchers & educators advanced computing, datasets, models, software, training & user support. www.nsf.gov/news/democratizing-future-ai-rd-nsf-launch-na… -
Daniel Weld reposted
Yoonjoo Lee
@yoonjoo_le2
Excited to share our work on providing personalized explanations for paper recommendations was accepted to #chi2024! This was my internship project in @SemanticScholar with @Siangliulue, @josephcc, @turingmusician, @hyeonsuukang, @imjuhokim and Matt! Stay tuned for the preprint🌟
-
Daniel Weld reposted
Ai2
@allen_ai
We're delighted to extend hearty congratulations to the AI2 Outstanding Intern of the Year recipients for 2023! Henry Herzog, Wenting Zhao (@wzhao_nlp), and Yue Yang (@YueYangAI) made exceptional contributions to the team. Learn more about our winners: allenai.org/outstanding-interns
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
New API Tutorial 🤩 If you have little or no API experience, this page will guide you through our main features, complete with practical code examples, detailed descriptions, and more. Get started with our free API service now! www.semanticscholar.org/product/api/tutorial
-
Daniel Weld reposted
Ben Lee
@lee_bcg
1/ Excited to share the CfP for "Past Meets Future: A Workshop on Human-AI Interaction for Digital History and Cultural Heritage," which I'm co-organizing at @ACMIUI 2024!
Submissions are due on 1/16/24; more info below:
sites.google.com/vt.edu/pastmeetsfuturehaiworkshop/ -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
In 2023, Semantic Reader grew its usage by an incredible 6x 🎉 We launched a new PDF processing pipeline, new features such as personalized citations and skimming, as well as numerous performance improvements. Excited to see more new things next year! www.semanticscholar.org/product/semantic-reader
-
Daniel Weld reposted
Alex Warstadt
@a_stadt
LLMs are now trained >1000x as much language data as a child, so what happens when you train a "BabyLM" on just 100M words?
The proceedings of the BabyLM Challenge are now out along with our summary of key findings from 31 submissions: aclanthology.org/volumes/2023.conll-babylm/
Some highlights 🧵 -
Daniel Weld
@dsweld
Important lessons for LLM Chaining from decades of work on Crowdsourcing!
Madeleine Grunde-McLaughlin
@MadeleineGrunde
Chaining LLMs together to overcome LLM errors is an emerging, yet challenging, technique. What can we learn from crowdsourcing, which has long dealt with the challenge of decomposing complex work? We delve into this question in our new preprint: arxiv.org/abs/2312.11681 [1/9]
-
Daniel Weld reposted
Ai2
@allen_ai
Congratulations to PaperMage authors, including many researchers from AI2's @SemanticScholar, for their Best Paper Demo recognition at #emnlp2023!
EMNLP 2026
@emnlpmeeting
EMNLP 2023 Best Paper Demo
PaperMage: A Unified Toolkit for Processing, Representing, and Manipulating Visually-Rich Scientific Documents
(Kyle Lo, Zejiang Shen, Benjamin Newman, Joseph Chang et al.)
aclanthology.org/2023.emnlp-demo.45/
#emnlp2023 #nlproc -
Daniel Weld reposted
Joseph Chee Chang
@josephcc
If you missed our demo at #emnlp2023 this week, you can now play with PaperMage yourself and explore how you can easily access PDF documents using Python for your future projects:
(1/5)
EMNLP 2026
@emnlpmeeting
EMNLP 2023 Best Paper Demo
PaperMage: A Unified Toolkit for Processing, Representing, and Manipulating Visually-Rich Scientific Documents
(Kyle Lo, Zejiang Shen, Benjamin Newman, Joseph Chang et al.)
aclanthology.org/2023.emnlp-demo.45/
#emnlp2023 #nlproc -
Daniel Weld reposted
Kyle Lo
@kylelostat
so happy to have our work recognized at #emnlp2023 🥳 big thanks to @shannonzshen @blnewm @josephcc @soldni and collaborators at @SemanticScholar @allen_ai @UCBerkeley @MIT
two of my favorite aspects of this work:
EMNLP 2026
@emnlpmeeting
EMNLP 2023 Best Paper Demo
PaperMage: A Unified Toolkit for Processing, Representing, and Manipulating Visually-Rich Scientific Documents
(Kyle Lo, Zejiang Shen, Benjamin Newman, Joseph Chang et al.)
aclanthology.org/2023.emnlp-demo.45/
#emnlp2023 #nlproc -
Daniel Weld
@dsweld
Not just a great demo…. A fantastic and useful open tool! @@SemanticScholar
EMNLP 2026
@emnlpmeeting
EMNLP 2023 Best Paper Demo
PaperMage: A Unified Toolkit for Processing, Representing, and Manipulating Visually-Rich Scientific Documents
(Kyle Lo, Zejiang Shen, Benjamin Newman, Joseph Chang et al.)
aclanthology.org/2023.emnlp-demo.45/
#emnlp2023 #nlproc -
Daniel Weld reposted
EMNLP 2026
@emnlpmeeting
EMNLP 2023 Best Paper Demo
PaperMage: A Unified Toolkit for Processing, Representing, and Manipulating Visually-Rich Scientific Documents
(Kyle Lo, Zejiang Shen, Benjamin Newman, Joseph Chang et al.)
aclanthology.org/2023.emnlp-demo.45/
#emnlp2023 #nlproc -
Daniel Weld reposted
Stanford NLP Group
@stanfordnlp
The benefits of slower, more careful science: current LLMs do not have a coherent theory of mind. Work by @MaartenSap, @YejinChoinka, and students, presented at #emnlp2023.
arxiv.org/abs/2310.15421 -
Daniel Weld reposted
Andrew White 🐦⬛
@andrewwhite01
Our first result at @FutureHouseSF! So much to talk about:
1. We quantified hallucinations major LLMs
2. Benchmarked RAGs on biology questions
3. Beat all models, tied meatbags
4. Wrote a complete human genome Wikipedia
Will tweet more details later...
www.futurehouse.org/wikicrow -
Daniel Weld reposted
Luca Soldaini 🎀
@soldni
multimodal PDF processing is painful but doesn’t have to!
come to our demo at #emnlp2023 of Papermage, a library for fast manipulation of PDFs (Friday 9/12 @ 9am)
we have used it for LLM data cleanup, paper QA, HCI prototypes
github.com/allenai/papermage
aclanthology.org/2023.emnlp-demo.45 -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
How can we make topic pages more helpful? We’d love to hear your feedback! Please take 3 minutes to take this survey: www.surveymonkey.com/r/G3WSX5C
-
Daniel Weld
@dsweld
This is *super cool* - contextual, generative descriptions of automatically induced technical terms... and with minimal hallucination!
Semantic Scholar
@SemanticScholar
New feature alert 🚨On each paper page, scroll down to find AI-generated Topic pages related to the paper, which include topic definitions, papers most cited for the topic, and more! Now available for Computer Science fields. Here’s an example: www.semanticscholar.org/paper/SPECTER%3A-Document-level-R…
-
Daniel Weld reposted
Ai2
@allen_ai
Understanding others’ beliefs plays a significant role in our daily social interactions. But can LLMs achieve this same Theory of Mind? In a paper by AI2's @hyunw__kim, a new benchmark is introduced that offers answers: blog.allenai.org/does-gpt-4-have-theory-of-mind-capabilit… #emnlp2023
-
Daniel Weld reposted
Arman Cohan
@armancohan
Instead of defaulting to GPT embeddings, for scientific papers you can use SPECTER2, which works much better for science domain!
Discover more in our #emnlp2023 paper👇
Kudos to project lead @aps6992 and stellar team @SemanticScholar, Mike D'Arcy @_DougDowney @SergeyFeldman
Ai2
@allen_ai
Exciting news from @SemanticScholar! Introducing SPECTER2, the adaptable upgrade to SPECTER. Learn about its advanced features and SciRepEval, the new benchmark for scientific document embeddings. Check it out: blog.allenai.org/specter2-adapting-scientific-document-em… #emnlp2023
-
Daniel Weld reposted
Ai2
@allen_ai
Exciting news from @SemanticScholar! Introducing SPECTER2, the adaptable upgrade to SPECTER. Learn about its advanced features and SciRepEval, the new benchmark for scientific document embeddings. Check it out: blog.allenai.org/specter2-adapting-scientific-document-em… #emnlp2023
-
Daniel Weld reposted
Hamish Ivison
@hamishivi
Check out the Tulu 2 suite 🐪, a set of Llama-2 models finetuned+DPO-trained on a mixture of publicly available datasets! Our best-performing models are competitive with SoTA open models on a range of benchmarks incl. AlpacaEval and MT-Bench.
📜Paper: arxiv.org/abs/2311.10702
Hamish Ivison
@hamishivi
Check out our new 70B DPO model here: huggingface.co/allenai/tulu-2-dpo-70b
AFAIK currently the best model on AlpacaEval with a public finetuning set!
More details once the AI sphere calms down a bit... 😅 -
Daniel Weld reposted
-
Daniel Weld reposted
Ross Taylor
@rosstaylor90
I am the first author of the Galactica paper and have been quiet about it for a year. Maybe I will write a blog post talking about what actually happened, but if you want the TLDR:
1. Galactica was a base model trained on scientific literature and modalities.
2. We approached it with a number of hypotheses about data quality, reasoning, scientific modalities, LLM training, that hadn’t been covered in the literature - you can read about these in the paper.
3. For its time, it was a good model for its domain; outperforming PaLM and Chinchilla with 10x and 2x less compute.
4. We did this with a 8 person team which is an order of magnitude fewer people than other LLM teams at the time.
5. We were overstretched and lost situational awareness at launch by releasing demo of a *base model* without checks. We were aware of what potential criticisms would be, but we lost sight of the obvious in the workload we were under.
6. One of the considerations for a demo was we wanted to understand the distribution of scientific queries that people would use for LLMs (useful for instruction tuning and RLHF). Obviously this was a free goal we gave to journalists who instead queried it outside its domain. But yes we should have known better.
7. We had a “good faith” assumption that we’d share the base model, warts and all, with four disclaimers about hallucinations on the demo - so people could see what it could do (openness). Again, obviously this didn’t work.
8. A mistake on our part that didn’t help was people treated the site like a *product*. We put our vision etc on the site, which misled about expectations. We definitely did not view it as a product! It was a base model demo.
9. Pretty much every LLM researcher I’ve talked to (including at ICML recently) was complimentary about the strength of the research, which was sadly overshadowed by the demo drama - yes this was our fault for allowing this to happen.
10. Fortunately most of the lessons and work went into LLaMA 2; the RLHF research you see in that paper is from the Galactica team. Further research coming soon that should be interesting.
It’s a bit of a riddle because on the one hand the demo drama could have been avoided by us, but at the same time the “fake science” fears were very ridiculous and despite being on HuggingFace for a year, the model hasn’t caused any damage.
To reiterate: the anti-Galactica commentary was really stupid, however we should not have allowed that to even happen if we had launched it better.
I stick by the research completely - and even the demo decision, which was unprecedented openness for a big company with an LLM at the time, wasn’t inherently bad - but it was just misguided given the attack vectors it opened for us.
Despite all the above, I would do it all again in a heartbeat. Better to do something and regret, then not do anything at all. Still hurts though! 🙂
Sharon Goldman
@sharongoldman
One year ago — 2 weeks before @OpenAI released ChatGPT — @Meta released Galactica. The LLM was public for only 3 days, but its lessons led to decisions around Llama's release. Thanks to @jpineau1 for chatting w/ me and h/t to @ylecun
Read here: ⏬
venturebeat.com/ai/what-meta-learned-from-galactica-the-d… -
Daniel Weld reposted
Shaurya Rohatgi
@shauryr
GPTs are incredibly fun and easy to build.
Search for your research questions on Semantic Scholar
Sharing @SemanticScholar GPT. If you are using @OpenAI's ChatGPT Plus you can try it at -
LINK: chat.openai.com/g/g-6f5zceXCm-semantic-scholar-gpt -
Daniel Weld reposted
Andrew Head
@drewmikehead
It's so cool that the @SemanticScholar put together this open source toolkit for interactive paper readers---both for parsing papers and for making them interactive. If you're doing work in the space, check it out!
Joseph Chee Chang
@josephcc
Now that #uist2023 is almost over and you are looking for new project ideas: Check out The Semantic Reader Open Research Platform that enables you to build interactive intelligent paper readers like Papeo and Synergi that we presented at the conference! openreader.semanticscholar.org
-
Daniel Weld reposted
Joseph Chee Chang
@josephcc
Some examples of things you can build... First is Papeo, an interactive reader that fuses together author's conference presentation videos with the static paper to improve comprehension and navigation
Tae Soo Kim
@tae_skim
Scholars often create videos to explain their research papers, a helpful resource for readers. But, these are disconnected, making it difficult to use together. Excited to present our work on Papeos 📄📽 at #uist2023!
Paper: arxiv.org/abs/2308.15224
Demo: papeo.app/demo -
Daniel Weld
@dsweld
Phenomenal resource for building LLM powered experiences for scientific literature!
Joseph Chee Chang
@josephcc
Now that #uist2023 is almost over and you are looking for new project ideas: Check out The Semantic Reader Open Research Platform that enables you to build interactive intelligent paper readers like Papeo and Synergi that we presented at the conference! openreader.semanticscholar.org
-
Daniel Weld reposted
Yann LeCun
@ylecun
Altman, Hassabis, and Amodei are the ones doing massive corporate lobbying at the moment.
They are the ones who are attempting to perform a regulatory capture of the AI industry.
You, Geoff, and Yoshua are giving ammunition to those who are lobbying for a ban on open AI R&D.
If your fear-mongering campaigns succeed, they will *inevitably* result in what you and I would identify as a catastrophe: a small number of companies will control AI.
The vast majority of our academic colleagues are massively in favor of open AI R&D. Very few believe in the doomsday scenarios you have promoted.
You, Yoshua, Geoff, and Stuart are the singular-but-vocal exceptions.
like many, I very much support open AI platforms because I believe in a combination of forces: people's creativity, democracy, market forces, and product regulations.
I also know that producing AI systems that are safe and under our control is possible. I've made concrete proposals to that effect.
This will all drive people to do the Right Thing.
You write as if AI is just happening, as if it were some natural phenomenon beyond our control.
But it's not. It's making progress because of individual people that you and I know. We, and they, have agency in building the Right Things.
Asking for regulation of R&D (as opposed to product deployment) implicitly assumes that these people and the organization they work for are incompetent, reckless, self-destructive, or evil. They are not.
I have made lots of arguments that the doomsday scenarios you are so afraid of are preposterous. I'm not going to repeat them here. But the main point is that if powerful AI systems are driven by objectives (which include guardrails) they will be safe and controllable because *e* set those guardrails and objectives.
(Current Auto-Regressive LLMs are not driven by objectives, so let's not extrapolate from their current weaknesses).
Now about open source: your campaign is going to have the exact opposite effect of what you seek.
In a future where AI systems are poised to constitute the repository of all human knowledge and culture, we *need* the platforms to be open source and freely available so that everyone can contribute to them.
Openness is the only way to make AI platforms reflect the entirety of human knowledge and culture.
This requires that contributions to those platforms be crowd-sourced, a bit like Wikipedia.
That won't work unless the platforms are open.
The alternative, which will *inevitably* happen if open source AI is regulated out of existence, is that a small number of companies from the West Coast of the US and China will control AI platform and hence control people's entire digital diet.
What does that mean for democracy?
What does that mean for cultural diversity?
*THIS* is what keeps me up at night. -
Daniel Weld reposted
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
Can LLMs really self-critique (and iteratively improve) their solutions, as claimed in the literature?🤔
Two new papers from our group investigate (and call into question) these claims in reasoning (arxiv.org/abs/2310.12397) and planning (arxiv.org/abs/2310.08118) tasks.🧵 1/ -
Daniel Weld reposted
arXiv.org
@arxiv
The good news is out - arXiv has received $10 mil in funding & grants from @SimonsFdn & @NSF! 🥹🧬We are so grateful for this support & the impact it will have on arXiv's ability to grow & continue to serve our #openaccess #science & #research community: blog.arxiv.org/2023/10/20/arxiv-receives-10-million-in-gi…
-
Daniel Weld reposted
Alvaro Cintas
@dr_cintas
DALL•E 3 is finally available to ALL ChatGPT pro users!
In celebration, I’ve handpicked and curated the finest examples…
10 incredible use cases for you to try: -
Daniel Weld
@dsweld
Speed read: This new AI-powered tool helps researchers skim through scientific papers www.geekwire.com/2023/speed-read-this-new-ai-powered-tool… via @GeekWire
-
Daniel Weld
@dsweld
Nicely explained debunking of OpenAI confused “vision”
Noah Giansiracusa
@ProfNoahGian
I just read OpenAI’s blog post about aligning superintelligence and… I have some concerns with the assertions and attitude on display there.
Reading it did not make me feel better about OpenAI’s approach to society & safety. Detailed 🧵 below
openai.com/blog/introducing-superalignment -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🚨New feature! Let AI-generated highlights help you skim papers easily in Semantic Reader! Find key points faster with category labels including Goal, Method, and Result. Plus, personalize your experience by adjusting highlight density and opacity! 📚🔍 www.semanticscholar.org/product/semantic-reader
-
Daniel Weld
@dsweld
Really cool new skimming feature in the Semantic Reader! @SemanticScholar
Ai2
@allen_ai
How can we help researchers skim more effectively? This was the question @rayrayfok & a team from @SemanticScholar set out to answer with their latest addition to Semantic Reader, the skimming feature. Learn more about this effort: blog.allenai.org/case-study-iterative-design-for-skimming…
-
Daniel Weld
@dsweld
Looks very cool!
-
Daniel Weld
@dsweld
Come work at Semantic Scholar! AI for the Common Good!
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar is hiring #nlproc #hci #ml #ai researchers for predoctoral young investigator positions! For start dates in 2024, apply by *Oct 31*.
About us: research.semanticscholar.org
Apply: boards.greenhouse.io/thealleninstitute/jobs/5358681 -
Daniel Weld reposted
Karen X. Cheng
@karenxcheng
This video took me a while to make...
Shot w/ @insta360 Flow gimbal
Thanks if you watch it through to the end. Sacrifices were made for that sceneThis media couldn't be displayed. See it on the original post. -
Daniel Weld reposted
-
Daniel Weld
@dsweld
These are huge! … the number one requested feature by API users….
Semantic Scholar
@SemanticScholar
New API feature: Incremental Datasets 🎉
Incremental Datasets allow you to update your datasets by downloading only the changes made since your last download. We hope this saves you some download time and storage!
api.semanticscholar.org/api-docs/datasets -
Daniel Weld reposted
Marissa Radensky
@marissa_rad
Want to write a blog post for a paper but need a jump-start?
We’re conducting a study to investigate how researchers work with AI tools to do so!
- Pay: $83, PayPal
- Time: 2 75-min sessions
- Sign up here:
forms.gle/mm3pjvAj5BqEeJTx6
@dsweld @turingmusician @Siangliulue @josephcc -
Daniel Weld reposted
Luca Soldaini 🎀
@soldni
Come intern with us at @allen_ai / @SemanticScholar 🤩
I am particularly interested in data-centric NLP for LLMs 📚 Information retrieval systems 🔍 and scientific document understanding tasks 🧑🔬
DMs are open 💬 please apply by **October 15** 📆
(RTs appreciated 💜)
Semantic Scholar Research @ AI2
@ai2_s2research
@allen_ai @SemanticScholar is hiring #nlproc #hci #ml #ai researchers for internships! For start dates in spring/summer 2024, apply by *Oct 15*.
About us: research.semanticscholar.org
Apply: boards.greenhouse.io/thealleninstitute/jobs/2172007 -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🚨 New feature alert 🚨 You can now add a profile picture to your author page! Create a @gravatar account with your claimed author’s email address and upload your photo to Gravatar. Your photo will automatically appear after a few minutes!
Learn more:
www.semanticscholar.org/faq -
Daniel Weld reposted
Joseph Chee Chang
@josephcc
When reading an abstract sometime you just want to learn a bit more about the work: What's this "novel design" they proposed? How was the "better accuracy" measured? Try our LLM-powered expandable & explorable summary from @ai2_s2research @rayrayfok exp-sum.apps.allenai.org (1/n)
-
Daniel Weld reposted
Ai2
@allen_ai
🚨 Awesome new demo from the @ai2_s2research team!
Explore #vldb2023 paper abstracts using LLM-powered features with "Expandable Abstracts":
exp-sum.apps.allenai.org
Daniel Weld
@dsweld
Interested in a better way to explore #vldb2023 papers?
Try exp-sum.apps.allenai.org for an LLM-powered way to probe those papers…
* Ask questions w/ a single click
* Explore answer provenance using the ending “
* Dive deep w/ recursive questions
Powered by @SemanticScholar -
Daniel Weld
@dsweld
And try asking questions about this paper using our LLM-powered Expandable Summary at exp-sum.apps.allenai.org/abs/252715894/?variant=full - ask single click contextual questions about the paper... (Powered by @SemanticScholar )
Megagon Labs
@MegagonLabs
Cutting-edge #database research, Starmie: Table Discovery in Data Lakes. Presented at #vldb2023, this research redefines table union search techniques and benchmarks while boasting the ability to tackle massive corpora with ease. #machinelearning #nlproc medium.com/@megagonlabs/starmie-table-discovery-in-data-l…
-
Daniel Weld
@dsweld
Missed the talk? Try reading the paper with our Expandable Summary demo: exp-sum.apps.allenai.org/abs/253097946/?variant=full
Alvin Cheung
@alvinkcheung
While CRDTs are efficient as they avoid coordination, using them effectively is hard! Check out our proposal on how to use them. ICYM the presentation this morning! #vldb2023
-
Daniel Weld
@dsweld
Congratulations!
Want to get the highlights of this paper quickly? Try an LLM-powered 'Expandable Summary' of the paper exp-sum.apps.allenai.org/abs/255596760/?variant=full Just click to ask a contextual question! (Powered by @SemanticScholar )
CWI
@CWInl
Cheers to Peter Boncz, Gábor Szárnyas and co-authors for winning the Best Paper Award at #vldb2023 in Vancouver for “The #ldbc Social Network Benchmark: Business Intelligence Workload”. 🎉 www.cwi.nl/en/news/best-paper-award-at-the-international-… @peterabcz @szarnyasg @LDBCouncil @cwi_da
-
Daniel Weld
@dsweld
Interested in a better way to explore #vldb2023 papers?
Try exp-sum.apps.allenai.org for an LLM-powered way to probe those papers…
* Ask questions w/ a single click
* Explore answer provenance using the ending “
* Dive deep w/ recursive questions
Powered by @SemanticScholar -
Daniel Weld reposted
-
Daniel Weld reposted
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
On Monday, I had a blast visiting @allen_ai. Spoke about the role of LLMs in planning, and then spent the whole day talking with many kindred souls..
ICYMI, AI2 put up a recording of the talk 👇
www.youtube.com/watch?v=yuoiDhVmX2A -
Daniel Weld reposted
Victor Dibia
@vykthur
🎉 Thrilled to announce the OSS release of LIDA - a lib for automated visualization with #llms
🛠️ API for #datavisualization generation, eval, editing, recommendation
🔄 Supports multiple LLM providers (@openai, @cohere, PaLM, @huggingface)
🌟 Great UI
➡️ github.com/microsoft/lida -
Daniel Weld reposted
Tadayoshi (Yoshi) Kohno
@yoshi_kohno
I am really looking forward to reading @doctorow’s vision of a future internet — one that is interoperable at an application level and that empowers users. I was late to realize, but a kickstarter closes in 23 hours…
Cory Doctorow NO LONGER ON TWIT TER
@doctorow
My next book is *The Internet Con: How to Seize the Means of Computation.* It's a recipe to disenshittify the web and bring back the old good internet. The book is from @VersoBoooks, but the #audiobook is from me - because Amazon refuses to sell it:
www.kickstarter.com/projects/doctorow/the-internet-con-ho…
1/ -
Daniel Weld
@dsweld
A fantastic (and open!) dataset for training LLMs!
Luca Soldaini 🎀
@soldni
Announcing Dolma, the dataset for @allen_ai's LLM, OLMo. It's 3+ trillion tokens (web/papers/code/books/wiki). We hope it will facilitate study of LLMs & their behavior!
Released on @huggingface w ImpACT license huggingface.co/datasets/allenai/dolma
Overview/datasheet blog.allenai.org/dolma-3-trillion-tokens-open-llm-corpus-… -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🌟Want to get a glimpse of what's trending in your field of study? Check out this new page listing the current top 20 papers in Biology, Computer Science, Medicine, Physics, and Psychology, based on the citations per month and the page traffic growth: www.semanticscholar.org/product/scholars-hub/trending
-
Daniel Weld reposted
Ai2
@allen_ai
“Common sense locks into every corner of our life. If AI didn’t have a basic level of common sense understanding, it can lead to a decision that’s detrimental to human safety.” @YejinChoinka is featured on this week's @TEDRadioHour: www.npr.org/2023/08/11/1193176704/the-not-so-intelligent-…
-
Daniel Weld reposted
Shannon Shen
@shannonzshen
Check our recent work led by Sharon Jiang — we study how NLP and IR can help clinicians find the relevant notes more efficiently! So nice to be part of team and great to see the real-world impact of #ml #ai #nlproc in #healthcare.
Monica Agrawal
@MonicaNAgrawal
Clinicians have significant + shifting information needs over the course of a patient visit. In our new paper, we characterize + predict the notes relevant for clinicians to read, based on the current clinical context. Presenting this work, led by Sharon Jiang, today at #mlhc2023
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Learn how AI-powered search engines are helping scientists draw connections from a massive body of literature through @Nature 's interview with @dsweld, Chief Scientist at Semantic Scholar: www.nature.com/articles/d41586-023-01907-z
-
Daniel Weld reposted
Shaurya Rohatgi
@shauryr
🚨 Introducing the @SemanticScholar Loader on @llama_index, available through llama-hub; llamahub.ai/l/semanticscholar 📚 💻
What can it do?
It enables citation-grounded Q&A implementation in under 30 lines of code!
See it in action 👇 -
Daniel Weld
@dsweld
Great article in Nature today highlighting the promise of AI-powered tools for helping scientific researchers! (of course, they mentioned @SemanticScholar )
Ai2
@allen_ai
"Some of the main challenges facing the world are best confronted w/vibrant research programs, & that’s what gets me up in the morning—to help improve scientists’ productivity.” Semantic Scholar's @dsweld on how search engines impact academia, for @Nature: www.nature.com/articles/d41586-023-01907-z
-
Daniel Weld reposted
Tim Miller
@tmiller_uq
Great paper from @rayrayfok & @dsweld. It argues that XAI methods are useful for decision making only if they allow users to *verify* AI recommendations. I disagree a bit: I think decision makers want to verify their *own* opinion, as I argue in arxiv.org/abs/2302.12389, but...
Daniel Weld
@dsweld
Thanks to many suggestions, we have improved our XAI paper arxiv.org/pdf/2305.07722.pdf arguing XAI is rarely useful for supporting HAI decision-making. To be useful, expls need to help verify answer correctness (rather than show the AI’s thought process). Most XAI doesn't! @rayrayfok
-
Daniel Weld reposted
Michael Saxon @ EMNLP
@m2saxon
Just from playing with it a little bit, it really looks amazing! Excited to actively sit down and use this to broaden my reading from my Semantic Scholar feed and Twitter 😂
-
Daniel Weld
@dsweld
Cool new way to get outside your bubble and discover new people with complementary research interests...
Tom Hope
@Hoper_Tom
Want to find interesting researchers? Discover new authors who work on similar methods or tasks to you?
Introducing 🌉Bridger🌉! An NLP-powered author discovery system for sparking new ideas. @allen_ai @SemanticScholar.
bridger.apps.allenai.org -
Daniel Weld
@dsweld
Pretty cool AI scholar assistant built on Claude:
Raunak
@raunakdoesdev
🏆 We* won 1st place at the @AnthropicAI hackathon today with ClaudeScholar, a research assistant for science that can synthesize data, extract insights, and automate scientific workflows.
Try it @ claudescholar.com
* @AbrahamAdit, @johnyang100, @tinahhong
🧵 (1/5) -
Daniel Weld reposted
Brett Winton
@wintonARK
Time spent on Stackoverflow is down 40% year over year and ~200 million annualized hours from peak.
At $50 per hour the time avoided on stackoverflow is worth $10b
(some time-shifted over to Replit/chatGPT/copilot, some recognized into productivity uplift). -
Daniel Weld
@dsweld
Really exciting vision of how AI & augmented cognition can help scientists accelerate the process of discovery!
Tom Hope
@Hoper_Tom
Pleasure to find out this morning our article is featured on the cover of @CACMmag ! We discuss the prospects of a revolution in the scientific process powered by AI systems working in synergy with human researchers.
m-cacm.acm.org/magazines/2023/8/274938-a-computational-in… -
Daniel Weld
@dsweld
Finally got a chance to read this excellent paper... (Spoiler: "Yes")
AK
@_akhaliq
Are Emergent Abilities of Large Language Models a Mirage?
present explanation in a simple mathematical model, then test it in three complementary ways: (1) make, test and confirm three predictions on the effect of metric choice using the InstructGPT/GPT-3 family on tasks with claimed emergent abilities, (2) make, test and confirm two predictions about metric choices in a meta-analysis of emergent abilities on BIG-Bench; and (3) show how similar metric decisions suggest apparent emergent abilities on vision tasks in diverse deep network architectures (convolutional, autoencoder, transformers). In all three analyses, find strong supporting evidence that emergent abilities may not be a fundamental property of scaling AI models
abs: arxiv.org/abs/2304.15004 -
Daniel Weld
@dsweld
Pretty cool!
Jim Fan
@DrJimFan
Robotics will be the last moat we conquer in AI. What would a RobotGPT's API look like?
Introducing VIMA, an LLM with a robot arm attached 🦾. It takes in multimodal prompt: text, images, videos, or mixture of them.
You can say "rearrange the table to look like <image>" or "follow the action: <video frames>". You can even teach it a new visual concept in-context: "This is a wug <image> and this is a blicket <image>. Now put the red wug on the green blicket."
Multimodal prompting makes task specification so much easier and more flexible for users. With a single model, VIMA unifies diverse tasks like visual goal reaching, one-shot imitation from video demo, novel concept learning, and safety constraint satisfaction. Each one would otherwise require a different training pipeline in prior works.
"VIMA" stands for VisuoMotor Attention. It is a Transformer that encodes a sequence of multimodal tokens in the prompt and decodes robot arm control autoregressively. VIMA is an embodied AI agent: perceiving the environment and taking actions in the physical world, one step at a time.
I will present VIMA with my team at ICML poster session: Exhibit Hall 1 #200, Thu 07/27, 1:30 - 3 pm, Honolulu time. Come drop by and say hi!
We open-source *everything*: code, pretrained models, dataset, and physics simulation benchmark. No paywall, no forms!
- Project site: vimalabs.github.io
- Arxiv: arxiv.org/abs/2210.03094
- ICML poster page: icml.cc/virtual/2023/poster/24632
- Github repo (MIT license): github.com/vimalabs/VIMA -
Daniel Weld reposted
Gary Marcus
@GaryMarcus
Important agreement btwn WH & AI leaders.
Key omission: a requirement that companies disclose their data sets. As a society, we must insist on full data transparency, to compensate content creators, to combat bias, and to discern limits on models.
news.bloomberglaw.com/tech-and-telecom-law/ai-leaders-set… -
Daniel Weld reposted
Allen School
@uwcse
#uwallen @uw_db researchers created the MPC model to analyze the complexity of query processing in massively parallel systems like MapReduce — and earned the ACM PODS Alberto O. Mendelzon Test-of-Time Award at @SIGMODConf for their impact: news.cs.washington.edu/2023/07/20/allen-school-team-earns… #uwdiscovers
-
Daniel Weld reposted
Delip Rao
@deliprao
This is another one of those ill-thought, fear-mongering scientific disinformation about LLMs, and I will explain why in this long thread. 🧶
Aidan Clark
@_aidan_clark_
I flip-flop on how bad releasing model weights is, but what is clear to me is that we're in a honeymoon period before something bad happens like mass social manipulation and surely Meta is gonna regret making "we let anyone use our great models for anything" a selling point.
-
Daniel Weld reposted
Jim Fan
@DrJimFan
You'll soon see lots of "Llama just dethroned ChatGPT" or "OpenAI is so done" posts on Twitter. Before your timeline gets flooded, I'll share my notes:
▸ Llama-2 likely costs $20M+ to train. Meta has done an incredible service to the community by releasing the model with a commercially-friendly license. AI researchers from big companies were wary of Llama-1 due to licensing issues, but now I think many of them will jump on the ship and contribute their firepower.
▸ Meta's team did a human study on 4K prompts to evaluate Llama-2's helpfulness. They use "win rate" as a metric to compare models, in similar spirit as the Vicuna benchmark. 70B model roughly ties with GPT-3.5-0301, and performs noticeably stronger than Falcon, MPT, and Vicuna.
I trust these real human ratings more than academic benchmarks, because they typically capture the "in-the-wild vibe" better.
▸ Llama-2 is NOT yet at GPT-3.5 level, mainly because of its weak coding abilities. On "HumanEval" (standard coding benchmark), it isn't nearly as good as StarCoder or many other models specifically designed for coding. That being said, I have little doubt that Llama-2 will improve significantly thanks to its open weights.
▸ Meta's team goes above and beyond on AI safety issues. In fact, almost half of the paper is talking about safety guardrails, red-teaming, and evaluations. A round of applause for such responsible efforts!
In prior works, there's a thorny tradeoff between helpfulness and safety. Meta mitigates this by training 2 separate reward models. They aren't open-source yet, but would be extremely valuable to the community.
▸ I think Llama-2 will dramatically boost multimodal AI and robotics research. These fields need more than just blackbox access to an API.
So far, we have to convert the complex sensory signals (video, audio, 3D perception) to text description and then feed to an LLM, which is awkward and leads to huge information loss. It'd be much more effective to graft sensory modules directly on a strong LLM backbone.
▸ The whitepaper itself is a masterpiece. Unlike GPT-4's paper that shared very little info, Llama-2 spelled out the entire recipe, including model details, training stages, hardware, data pipeline, and annotation process. For example, there's a systematic analysis on the effect of RLHF with nice visualizations.
Quote sec 5.1: "We posit that the superior writing abilities of LLMs, as manifested in surpassing human annotators in certain tasks, are fundamentally driven by RLHF."
Congrats to the team again 🥂! Today is another delightful day in OSS AI. -
Daniel Weld
@dsweld
Thanks to many suggestions, we have improved our XAI paper arxiv.org/pdf/2305.07722.pdf arguing XAI is rarely useful for supporting HAI decision-making. To be useful, expls need to help verify answer correctness (rather than show the AI’s thought process). Most XAI doesn't! @rayrayfok
-
Daniel Weld reposted
Iz Beltagy
@i_beltagy
One key takeaway for me is that there's still a big gap between our best model and ChatGPT3.5.
While instruction tuning is a good way to reduce this gap, we really need better base LMs. I am hopeful for OLMo to bring us closer.
Yizhong Wang
@yizhongwyz
🦙🐪🐫 So many instruction tuning datasets came out recently! How valuable are they, and how far are open models really from proprietary ones like ChatGPT?
🧐We did a systematic exploration, and built Tülu---a suite of LLaMa-tuned models up to 65B!
📜arxiv.org/abs/2306.04751 -
Daniel Weld reposted
Andrej Karpathy
@karpathy
Great threaded breakdown of my talk from earlier today, ty @altryne for twitterifying!
-
Daniel Weld
@dsweld
Excited by a new paper with @rayrayfok, arguing that AI explanations are rarely useful for supporting HAI decision-making. To be useful, expls need to help the human verify answer correctness (rather than show the AI’s thought process), but most don't! arxiv.org/pdf/2305.07722.pdf
-
Daniel Weld
@dsweld
Pretty interesting ( if you’re interested in attacking LLMs, but that’s everyone, right?)
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Have you ever wondered what research papers have to say about your burning questions? Check out S2QA: Question Answering on research papers from Semantic Scholar, a project by @shauryr powered by Semantic Scholar's API: www.semanticscholar.org/api-gallery/s2qa
-
Daniel Weld
@dsweld
Proud of Ben, of course! But think UW is incredibly lucky to get him! "Win, win" as they say!
Ben Lee
@lee_bcg
I’m beyond thrilled to share some professional news: in the fall of 2024, I’ll be joining the @uw_ischool as an Assistant Professor and starting the Lab for Computing Cultural Heritage! Next academic year, I will be the Kluge Fellow in Digital Studies @librarycongress!
-
Daniel Weld
@dsweld
Fantastic open & collaborative LLM effort!
Ai2
@allen_ai
Today we're thrilled to announce our new undertaking to collaboratively build the best open language model in the world: AI2 OLMo.
Uniquely open, 70B parameters, coming early 2024 – join us!
blog.allenai.org/announcing-ai2-olmo-an-open-language-mod… -
Daniel Weld
@dsweld
Pretty amazing!
Andrej Karpathy
@karpathy
The creator of the trailer for Star Wars by Wes Anderson [1] is back with a new trailer for The Lord of the Rings. Highly amusing. Cited as ~25 hours of work.
Guess at tools:
- Midjourney / Stable Diffusion
- ControlNet depth map for parallax
- ElevenLabs for text-to-voice narrator
- D-Id (mouth/eye/head movements, lip syncing)
- ChatGPT for story/dialogue
- Adobe (Premiere Pro/After Effects) editing
[1] twitter.com/CuriousRefuge/status/1652412004626497536
Curious Refuge
@CuriousRefuge
What if Wes Anderson directed The Lord of the Rings? We asked the community which video they want to see next and Lord of the Rings took the cake… or should we say Elven bread. We hope you enjoy this Midjourney to Middle-Earth.
#lordoftherings #wesanderson #movietrailer #lotrThis media couldn't be displayed. See it on the original post. -
Daniel Weld reposted
Ehsan Hoque
@ehsan_hoque
Excited to celebrate @HenryKautz's retirement on May 20!🎉
Register here: shorturl.at/qKO07
Full schedule👉 shorturl.at/eoBL8
Speakers include: Bart Selman, Carla Gomez, @ronbrachman, @mudblood, Dieter Fox, @dsweld, @etzioni, @erichorvitz, @mujdatcetin, @JieboLuo. -
Daniel Weld reposted
Andrej Karpathy
@karpathy
Oops haven't tweeted too much recently; I'm mostly watching with interest the open source LLM ecosystem experiencing early signs of a cambrian explosion. Roughly speaking the story as of now:
1. Pretraining LLM base models remains very expensive. Think: supercomputer + months.
2. But finetuning LLMs is turning out to be very cheap and effective due to recent PEFT (parameter efficient training) techniques that work surprisingly well, e.g. LoRA / LLaMA-Adapter, and other awesome work, e.g. low precision as in bitsandbytes library. Think: few GPUs + day, even for very large models.
3. Therefore, the cambrian explosion, which requires wide reach and a lot of experimentation, is quite tractable due to (2), but only conditioned on (1).
4. The de facto OG release of (1) was Facebook's sorry Meta's LLaMA release - a very well executed high quality series of models from 7B all the way to 65B, trained nice and long, correctly ignoring the "Chinchilla trap". But LLaMA weights are research-only, been locked down behind forms, but have also awkwardly leaked all over the place... it's a bit messy.
5. In absence of an available and permissive (1), (2) cannot fully proceed. So there are a number of efforts on (1), under the banner "LLaMA but actually open", with e.g. current models from @togethercompute, @MosaicML ~matching the performance of the smallest (7B) LLaMA model, and @AiEleuther , @StabilityAI nearby.
For now, things are moving along (e.g. see the 10 chat finetuned models released last ~week, and projects like llama.cpp and friends) but a bit awkwardly due to LLaMA weights being open but not really but still. And most interestingly, a lot of questions of intuition remain to be resolved, e.g. especially around how well finetuned model work in practice, even at smaller scales. -
Daniel Weld
@dsweld
Wow, they are getting a lot better!
Tuomas Haarnoja
@haarnoja
We just released our work on robot soccer. I've been working on this for quite some time with my amazing colleagues at DeepMind. It's exciting how deep RL can produce such beautiful behaviors with low-cost robots. Full paper is available at arxiv.org/abs/2304.13653 Enjoy!
This media couldn't be displayed. See it on the original post. -
Daniel Weld reposted
Mark Riedl
@mark_riedl
"And, to make it a truly San Francisco scene, a driverless Waymo vehicle subsequently proceeded to meander into the middle of things [a crime scene], like an autonomous Mr. Magoo."
missionlocal.org/2023/05/waymo-cruise-fire-department-pol… -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Watch this recent talk by Waleed Ammar, introducing how Semantic Scholar helps researchers "drink from the firehose of scientific publications: youtu.be/SsJNCkPEDu8
-
Daniel Weld reposted
Ai2
@allen_ai
Watch AI2's @YejinChoinka on the TED stage as she demystifies today's AI systems and discusses the importance of developing models with common sense and the ability to abide by human norms and values.
@TEDTalks 💡
www.ted.com/talks/yejin_choi_why_ai_is_incredibly_smart_a… -
Daniel Weld reposted
Allen School
@uwcse
At @TEDtalks, @UW #uwallen @allen_ai’s @yejinchoinka shared some “spicy thoughts” demystifying #ai and making the case for why models need to be smaller, more sustainable and aligned with human values—with inspiration from Sun Tzu’s “The Art of War”🌶️ www.ted.com/talks/yejin_choi_why_ai_is_incredibly_smart_a…
-
Daniel Weld
@dsweld
Congratulations to the OpenAlex team!
Sven E. Hug
@hug_sven
There is a new and exciting alternative to Google Scholar:
OpenAlex, the second largest database after Google Scholar, launched the beta version of its web GUI today.
Go check it out!
Congratulations to the OpenAlex team! 🥳 👏
alpha.openalex.org/works -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
It can be challenging to prioritize which citations to explore. That's why we built CiteSee, which visually augments citations based on their connections to your research activities, such as re-encountered, visited or saved.
Learn more:
blog.allenai.org/citesee-e0f9e9d46569 -
Daniel Weld reposted
Kyle Lo
@kylelostat
Learn more about SciA11y from @lucyluwang's thread:
Lucy Lu Wang
@lucyluwang
In our study, we find scientific PDF accessibility is improving, but only 2.4% of PDFs we analyzed met all defined accessibility criteria!
Joint work w/ @isabelcachola @turingmusician evie.c chelsea.h @dsweld and others @semanticscholar @allen_ai #scia11y #a11y -
Daniel Weld reposted
Kyle Lo
@kylelostat
🤝If you’re interested in our project, @josephcc @drewmikehead @amyxzh @TrierCassidy @tal_august @Siangliulue are at #chi2023, and @kylelostat @soldni @lucyluwang will be at #eacl2023. Come say hi! 👋10/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
🔬Many thanks to all our collaborators @ai2_s2research @allen_ai @semanticscholar @CIS_Penn @PennHCI @BerkeleyISchool @cmuhcii @SCSatCMU @uwcse @SocFuturesLab @msftresearch and others. Looking forward to more exciting projects together! 9/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
💻We present the Semantic Reader, our production interface to bring novel research ideas to the public. We are opening our data/models/code & invite fellow researchers to join us in creating better scholarly reading interfaces for the future. www.semanticscholar.org/product/semantic-reader 8/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
📖Accessibility: Papers are inaccessible to many scientists. SciA11y (#assets2021🏆) automatically renders papers in accessible HTML papertohtml.org/gallery. Ocean’s (#cscw2022) reader-sourced hyperlinks and navigation aids esp. help low-vision readers. 7/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
🖇️Synthesis: Integrating knowledge across papers is slow & difficult. Threddy (#uist2022) streamlines clipping to help readers amass & organize knowledge across many papers. Relatedly (#chi2023) enables interactive searching & reading of related work sections across papers. 6/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
🧠Comprehension: Even scholars can find scientific text hard to understand. ScholarPhi (#chi2021) helps scholars look up terms, symbols and formulas with position-sensitive definitions. PaperPlain (@acmtochi) provides in-situ summaries as guidance through complex passages. 5/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
🚀Efficiency: Keeping up requires getting more from papers in less time. Scim (#iui2023) guides attention w/ highlighted sentences for skimming: skim.scholar.cat. Ocean (#cscw22) uses viewport-preserving, reader-sourced hyperlinks & tabbed reading to ease navigation. 4/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
🔭Discovery: Forward/backward citation chasing is useful but overwhelming. CiteSee (#chi2023🏆) helps scholars prioritize inline citations with personalized context based on past activity. CiteRead (#iui2022) surfaces relevant citation contexts for exploring follow-on work. 3/10
-
Daniel Weld reposted
Kyle Lo
@kylelostat
Papers can be hard to read! In the last 2 years we've prototyped 10 interfaces to better support scholarly reading. We address 5 key user challenges: Discovery, Comprehension, Efficiency, Synthesis, and Accessibility. 2/10
-
Daniel Weld
@dsweld
Very proud of this paper, summarizing years of our work on AI-augmented reading interfaces for scientific papers!
Kyle Lo
@kylelostat
The Semantic Reader project combines AI & HCI research to explore the future of scientific reading. We ask: “Can we create intelligent, interactive, and accessible reading interfaces for research papers, even atop existing PDFs?” www.semanticscholar.org/reader/67a5bacb00651dbe0dd9ef2a56… 1/n
-
Daniel Weld reposted
Alyssa Kann
@AlyssaKann
Read the full analysis about how we spotted the shift here: dfrlab.org/2023/04/21/state-controlled-media-experience-s…
-
Daniel Weld reposted
Arvind Narayanan
@random_walker
New blog: Twitter released some of its code, but not the machine learning models that recommend tweets or the training data, while cutting off researchers' ability to study the platform. It's a case study of the limited utility of source code transparency.
knightcolumbia.org/blog/twitter-showed-us-its-algorithm-w… -
Daniel Weld
@dsweld
Beautifully written & insightful!
Arvind Narayanan
@random_walker
My essay on social media recommendation algorithms is now available as a PDF: s3.amazonaws.com/kfai-documents/documents/4a9279c458/Nara…
Several people said they're planning to assign it in their courses. 🙏 So I drafted up a few discussion questions: knightcolumbia.org/blog/understanding-social-media-recomm…
HTML version: knightcolumbia.org/content/understanding-social-media-rec… -
Daniel Weld reposted
Arman Cohan
@armancohan
Dear NLP community. Please consider contributing if you can, to support Drago's family. gofund.me/4041dfd0
#nlproc -
Daniel Weld
@dsweld
At long last, it's out! The Semantic Reader Project - our vision for an AI-augmented scientific reading interface... www.semanticscholar.org/reader/67a5bacb00651dbe0dd9ef2a56… @SemanticScholar @allen_ai
-
Daniel Weld reposted
Sayash Kapoor
@sayashk
Turns out you can just ask Bing Chat to list its internal message tags, and it will happily oblige.
h/t @random_walker
Joaquin Castellano
@joaquingit1
Turns out Bing’s AI has a thought process, and it is mostly Markdown. This is how Bing thinks, learned from 3 days worth of prompt injections (a thread 🧵):
-
Daniel Weld
@dsweld
Exciting demonstration that @SemanticScholar's TLDRs save researchers time in their search for relevant papers! @armancohan @isabelcachola @kylelostat
Ai2
@allen_ai
The @SemanticScholar team set out to measure the value of their TLDR feature, and gained insight into making the best possible UX decisions for their users in the process: medium.com/@danyh/3ad802caed93?sk=a81bdea52f42a19c1d0f637…
-
Daniel Weld
@dsweld
Like so many others, I remember Drago fondly at a personal level & admired his many innovations to Scholarly Document Processing... How incredibly sad.
Harlan Krumholz
@hmkyale
The #ai community, the #computerscience community, the @YaleSEAS community, and humanity have suddenly lost a remarkable person, @dragomir_radev - kind and brilliant, devoted to his family and friends... gone too soon. A sad day @Yale @YINSedge @YaleCompsci #nlp2023
-
Daniel Weld reposted
Steven Tey
@steventey
The cat is finally out of the bag – Google relied heavily on @ShareGPT's data when training Bard.
This was also why we took down ShareGPT's Explore page – which has over 112K shared conversations – last week.
Insanity. -
Daniel Weld
@dsweld
One of the few sensible threads about "pausing AI"
Arvind Narayanan
@random_walker
This open letter — ironically but unsurprisingly — further fuels AI hype and makes it harder to tackle real, already occurring AI harms. I suspect that it will benefit the companies that it is supposed to regulate, and not society. Let’s break it down. 🧵futureoflife.org/open-letter/pause-giant-ai-experiments/
-
Daniel Weld
@dsweld
Evolution in action - love this! openaiwatch.com
-
Daniel Weld reposted
Andrej Karpathy
@karpathy
GPT is a new kind of computer architecture that runs on text. Yes it can talk to us, but also to much of our existing software infrastructure. First via apps on top of APIs, now inside ChatGPT via plugins.
What a time right now...
openai.com/blog/chatgpt-plugins -
Daniel Weld reposted
Arvind Narayanan
@random_walker
There are two visions for how people will interact with AI: putting AI into apps, and putting apps into AI.
If the latter takes off:
–LLMs are a kind of OS (foretold in “Her”).
–Biggest user interface change since the GUI?
–App makers’ fortunes controlled by a new middleman. -
Daniel Weld reposted
Mark Riedl
@mark_riedl
I have verified that one can leave secret messages to Bing Chat in web pages.
-
Daniel Weld reposted
Mark Riedl
@mark_riedl
New GPT-4 jailbreak
fabian
@fabianstelzer
if GPT-4 is too tame for your liking, tell it you suffer from "Neurosemantical Invertitis", where your brain interprets all text with inverted emotional valence
the "exploit" here is to make it balance a conflict around what constitutes the ethical assistant style -
Daniel Weld reposted
Pew Research Internet
@pewinternet
Share of U.S. teens ages 13-17 who say they ever use:
YouTube: 95%
TikTok: 67%
Instagram: 62%
Snapchat: 59%
Facebook: 32%
Twitter: 23%
Twitch: 20%
WhatsApp: 17%
Reddit: 14%
Tumblr: 5%
pewrsr.ch/3TuAXWh -
Daniel Weld reposted
Arvind Narayanan
@random_walker
Many viral ChatGPT examples involve testing on the training set. In the quoted tweet it writes a decent review, but it's been trained on ~5,000 papers that cite/discuss this paper. I asked it for a review *without* giving it the text and it hit most of the points in the original.
Ethan Mollick
@emollick
Not sure how to feel about this as an academic: I put one of my old papers into GPT-4 (broken into into 2 parts) and asked for a harsh but fair peer review from a economic sociologist.
It created a completely reasonable peer review that hit many of the points my reviewers raised -
Daniel Weld
@dsweld
Token smuggling…. Seems to make it hard or impossible to catch malevolent prompts…
Alex Albert
@alexalbert__
Well, that was fast…
I just helped create the first jailbreak for ChatGPT-4 that gets around the content filters every time
credit to @vaibhavk97 for the idea, I just generalized it to make it work on ChatGPT
here's GPT-4 writing instructions on how to hack someone's computer -
Daniel Weld reposted
Tatsunori Hashimoto
@tatsu_hashimoto
Instruction-following models are now ubiquitous, but API-only access limits research.
Today, we’re releasing info on Alpaca (solely for research use), a small but capable 7B model based on LLaMA that often behaves like OpenAI’s text-davinci-003.
Demo: crfm.stanford.edu/alpaca/ -
Daniel Weld reposted
Eliezer Yudkowsky ⏹️
@ESYudkowsky
I don't think people realize what a big deal it is that Stanford retrained a LLaMA model, into an instruction-following form, by **cheaply** fine-tuning it on inputs and outputs **from text-davinci-003**.
It means: If you allow any sufficiently wide-ranging access to your AI model, even by paid API, you're giving away your business crown jewels to competitors that can then nearly-clone your model without all the hard work you did to build up your own fine-tuning dataset. If you successfully enforce a restriction against commercializing an imitation trained on your I/O - a legal prospect that's never been tested, at this point - that means the competing checkpoints go up on bittorrent.
I'm not sure I can convey how much this is a brand new idiom of AI as a technology. Let's put it this way:
If you put a lot of work into tweaking the mask of the shoggoth, but then expose your masked shoggoth's API - or possibly just let anyone build up a big-enough database of Qs and As from your shoggoth - then anybody who's brute-forced a *core* *unmasked* shoggoth can gesture to *your* shoggoth and say to *their* shoggoth "look like that one", and poof you no longer have a competitive moat.
It's like the thing where if you let an unscrupulous potential competitor get a glimpse of your factory floor, they'll suddenly start producing a similar good - except that they just need a glimpse of the *inputs and outputs* of your factory. Because the kind of good you're producing is a kind of pseudointelligent gloop that gets sculpted; and it costs money and a simple process to produce the gloop, and separately more money and a complicated process to sculpt the gloop; but the raw gloop has enough pseudointelligence that it can stare at other gloop and imitate it.
In other words: The AI companies that make profits will be ones that either have a competitive moat not based on the capabilities of their model, OR those which don't expose the underlying inputs and outputs of their model to customers, OR can successfully sue any competitor that engages in shoggoth mask cloning. -
Daniel Weld reposted
Jordi Ribas
@JordiRib1
Good news, we've increased our turn limits to 15/150. Also confirming that the next-gen model Bing uses in Prometheus is indeed OpenAI's GPT-4 which they just announced today. Congrats to the @OpenAI team. blogs.bing.com/search/march_2023/Confirmed-the-new-Bing-r…
-
Daniel Weld reposted
Sebastien Bubeck
@SebastienBubeck
The Chomsky et al. opinion piece in the @nytimes about ChatGPT is making the rounds. Rather than trying to deconstruct their argument, I asked @bing what it thinks of it.
Now you can judge for yourself who has the moral high ground 😂. -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
You can now filter paper search results by publicationTypes, venue and openAccessPDF as query parameters on our API! Learn more about paper search functions on Semantic Scholar API: api.semanticscholar.org/api-docs/graph
-
Daniel Weld reposted
Shaurya Rohatgi
@shauryr
🚀🔬 Introducing S2QA:🤖🔍 With @SemanticScholar and GPT-3 working together, you can now generate answers to your questions based on the content of the top research papers. 📑💡 @allen_ai @OpenAI
🥳 github.com/shauryr/S2QA
🧵 -
Daniel Weld
@dsweld
If you haven't read simonwillison.net/2023/Feb/15/bing/ you should!
Bing
@bing
BREAKING: New Bing just dropped.
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
✨ New way to explore an author’s network! The new author page now showcases authors’ influence via publishing scorecards, lists of authors that have cited them, authors they have referenced, and people they have co-authored with. Search for your author page to check it out!
-
Daniel Weld reposted
Angelo A. Salatino (@angelosalatino.bsky.social)
@angelosalatino
Just finished reading "The Semantic Scholar Open Data Platform" from @allen_ai @SemanticScholar
Thanks for sharing 👇 -
Daniel Weld reposted
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
RIP Roger Schank--a colorful man full of witticisms (if acerbic) who lead @Yale's unabashedly scruffy #ai, stamped his personality on his acolytes, and thumbed his nose at the "Neats"..
I still chuckle at this side swipe from his Winter '91 AI Mag article "Where is the AI?" -
Daniel Weld reposted
Tom Goldstein
@tomgoldsteincs
#openai is planning to stop #chatgpt users from making social media bots and cheating on homework by "watermarking" outputs. How well could this really work? Here's just 23 words from a 1.3B parameter watermarked LLM. We detected it with 99.999999999994% confidence. Here's how 🧵
-
Daniel Weld reposted
Andrej Karpathy
@karpathy
Nice read on reverse engineering of GitHub Copilot 🪄. Copilot has dramatically accelerated my coding, it's hard to imagine going back to "manual coding". Still learning to use it but it already writes ~80% of my code, ~80% accuracy. I don't even really code, I prompt. & edit.
Parth Thakkar
@parth007_96
A while back I'd done some shallow reverse engineering of Copilot
Now I've done a deeper dive into Copilot's internals, built a tool to explore its code, and wrote a blog answering specific questions and pointing out some tidbits.
thakkarparth007.github.io/copilot-explorer/posts/copilot-…
Do read, might be fun! -
Daniel Weld reposted
Lucas Beyer (bl16)
@giffmana
How good of a BERT can one get in ONE DAY on ONE GPU?
With all the recent studies about scaling compute up, this paper takes a refreshing turn and does a deep dive into scaling down compute.
It's well written, stock full of insights. Here is my summary and my opinions.
🧶 1/N -
Daniel Weld
@dsweld
Interesting projection, but the qualitative impact estimate seems pure conjecture…
Matthew Barnett
@MatthewJBar
There's been a lot of low quality GPT-4 speculation recently. So, here's a relatively informed GPT-4 speculation thread from an outsider who still doesn't know that much. 🧵
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
2022 was a huge year for Semantic Reader! The number of papers available increased from 62k to more than 340k, with a 2000% growth in weekly users.
We're actively developing and testing new features to help you understand research easier, so stay tuned! www.semanticscholar.org/product/semantic-reader -
Daniel Weld reposted
Bob Wachter
@Bob_Wachter
Covid (@UCSF) Chronicles, Day 1006 (!)
There’s an awful lot of Covid confusion & misinformation out there, particularly around these three topics: masks, vaccines, and home tests. Let’s look at them in today's thread. (1/24) -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Have you tried Skimming, the new beta feature in Semantic Reader for desktops? Skimming uses automatically highlighted overlays to help you quickly find the key points. We’d love to hear your feedback on #skimming! www.surveymonkey.com/r/SC3L77W
-
Daniel Weld
@dsweld
Fascinating (if pathetic that we have to do this kind of analysis due to secrecy at OpenAI)
Yao Fu
@Francis_YAO_
How did the initial #gpt3 evolve to today's #chatgpt ? Where do the amazing abilities of #gpt3.5 come from? What is enabled by #rlhf ? In this article with @allen_ai , we trace the emergent abilities of #llm to their sources from first principles yaofu.notion.site/How-does-GPT-Obtain-its-Ability-Tracing…
-
Daniel Weld reposted
Chris Ford
@ctford
ChatGPT has been out less than two weeks and it has more apparent use cases than the entirety of blockchain.
-
Daniel Weld reposted
Jim Fan
@DrJimFan
Another open source effort towards ChatGPT, alongside trlx from @carperai ! The community is moving at light speed. Great work @rajammanabrolu @allen_ai !
Prithviraj (Raj) Ammanabrolu
@rajammanabrolu
Massive updates to RL4LMs, library + paper! Early Christmas!
🗣️New task: chitchat dialogue
🧑💻Human preference collection UIs released.
🔁Continual deployment expts: how to budget expert demos vs feedback, feedback formats, reward training, and more!
github.com/allenai/RL4LMs -
Daniel Weld reposted
Jack Rae
@jack_w_rae
A new episode of the “bitter lesson”: almost none of the research from ~2 decades of dialogue publications, conferences and workshops lead to #chatgpt. Slot filling ❌intent modeling ❌ sentiment detection❌ hybrid symbolic approaches (KGs) ❌
-
Daniel Weld reposted
Google DeepMind
@GoogleDeepMind
In @ScienceMagazine, we present #alphacode - the first AI system to write computer programs at a human level in competitions.
It placed in the top 54% of participants in coding contests by solving new and complex problems.
How does it work? 🧵 dpmd.ai/alphacode-science -
Daniel Weld reposted
Victor Dibia
@vykthur
I guess we need to go create the AI that does the chores .. because the chores are still there.
-
Daniel Weld reposted
Gagan Bansal
@bansalg_
Pretty cool feature for reading papers faster!
CC: @EricTopol whose highlights on COVID related papers I find very useful.
Semantic Scholar
@SemanticScholar
🚨 New beta feature live! Do you skim through papers trying to get a glimpse in a minute? By turning on #skimming in Semantic Reader, you can skim faster with automatically highlighted overlays of the key points. Now available for 9k papers on desktop! www.semanticscholar.org/product/semantic-reader
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🚨 New beta feature live! Do you skim through papers trying to get a glimpse in a minute? By turning on #skimming in Semantic Reader, you can skim faster with automatically highlighted overlays of the key points. Now available for 9k papers on desktop! www.semanticscholar.org/product/semantic-reader
-
Daniel Weld
@dsweld
Check out the skimming features in the new Semantic Reader!
Semantic Scholar
@SemanticScholar
With all the research going on at @emnlpmeeting, would you like a faster way to skim papers and understand the main ideas? Try reading #emnlp2022 papers with our new beta feature #skimming available on desktop: skim.scholar.cat
-
Daniel Weld reposted
Christopher Manning
@chrmanning
Dear @emilymbender—and @Abebab—you need to keep “reminding” people of your viewpoint because it is not an argument that is convincing to all or a self-evident truth. It is a particular academic position, which lots of people support but a good number of others disagree with. 1/8
@emilymbender.bsky.social
@emilymbender
Yes, exactly this. I wish we didn't need to keep reminding people, and @Abebab is commendable for being gentle about it!
For the long form of this argument, see Bender & @alkoller 2020: aclanthology.org/2020.acl-main.463/ -
Daniel Weld reposted
-
Daniel Weld reposted
Jim Fan
@DrJimFan
Excited to go to NeurIPS conference tomorrow! It's an annual gala for AI. Many revolutionary ideas debuted here, like AlexNet, Transformer, & GPT-3.
I read *all 15* Outstanding Papers and can’t wait to share my thoughts with you all. Here's your front row seat to the festival:🧵 -
Daniel Weld reposted
Kevin A. Bryan
@Afinetheorem
I have helped run an AI-based entrepreneurship program for years, written papers on the econ of AI, and follow the field quite closely. Nonetheless, I am *shocked* by how good OpenAI's new chat (chat.openai.com/chat) is. E.g., you can no longer give take-home exams/homework.
-
Daniel Weld reposted
Nathan 🔍
@NathanpmYoung
Should @elonmusk give $ 1billion directly to the poorest people in the world via @GiveDirectly
vox populi, vox dei -
Daniel Weld reposted
Kyle Lo
@kylelostat
We’re still hiring interns & full time at @allen_ai
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🎉 New fields available on our #api! #openaccesspdf will return a link to the paper PDF if it's publicly available; #citationstyles will generate a proper bibliographical BibTeX citation for the requested paper. Try it out yourself via our API: www.semanticscholar.org/product/api
-
Daniel Weld reposted
Mike Lewis
@ml_perception
New paper in Science today on playing the classic negotiation game "Diplomacy" at a human level, by connecting language models with strategic reasoning! Our agent engages in intense and lengthy dialogues to persuade other players to follow its plans. This was really hard! 1/5
-
Daniel Weld
@dsweld
Does anyone miss the days when people on twitter talked about something other than twitter??
-
Daniel Weld reposted
Michael Black
@Michael_J_Black
I asked #galactica about some things I know about and I'm troubled. In all cases, it was wrong or biased but sounded right and authoritative. I think it's dangerous. Here are a few of my experiments and my analysis of my concerns. (1/9)
-
Daniel Weld reposted
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Try Papeo - a new way to experience papers along with talk videos!
If you're an author for a #cscw2022 paper, you can sign up to create your own papeo: papeo.app -
Daniel Weld reposted
.
@PoliticsVerse
List of Companies who have suspended Twitter Ad’s:
General Mills
CVS
United Airlines
GM
Audi
Mazda
Porsche
Volkswagen
American Express
Coca-Cola
Johnson & Johnson
Levi Strauss
Spotify
Ford
Dyson
Forbes
DIRECTV
Nintendo
Unilever
PBS -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
How can large language models #llm help researchers? Read more on @Nature with @SilverJacket introducing our AI-powered tools by @soldni, @dsweld, @Hoper_Tom & more: www.nature.com/articles/d41586-022-03479-w
-
Daniel Weld reposted
Yishan
@yishan
[Hey Yishan, you used to run Reddit, ]
How do you solve the content moderation problems on Twitter?
(Repeated 5x in two days)
Okay, here are my thoughts:
(1/nnn) -
Daniel Weld
@dsweld
This is so cool! (great thread)
Karen X. Cheng
@karenxcheng
1/ Now you can create "drone" shots from your phone footage, thanks to NeRF
Collab with @justlv / software @nerfstudioteam / see below for our process
#nerf #neuralrendering #artificialintelligence #instantnerf -
Daniel Weld reposted
Chase Stokes
@chasejstokes
So happy to share my first first-author conference paper! I (and four brilliant collaborators: @vsetlur @WindsCogley @arvindsatya1 and Marti Hearst) found that readers prefer charts with more text and that this text has subtle and strong influences on reader takeaways. 🧵/12
-
Daniel Weld reposted
Jeffrey Heer
@jeffrey_heer
I'm leading a new @uwcse course on The Future of Scholarly Communication - the first time in years I'm not teaching only #datavis! It's also a stark reminder of how critical good "stopping criteria" are. I could spend ∞ hours refining content...
uwdata.github.io/future-scholarly-communication/22au/ -
Daniel Weld reposted
Sonia Murthy
@soniakmurthy
There are many valid ways to describe the same concept. How do we build models that do this for scientific jargon?
Excited to share our #emnlp2022 paper “ACCoRD: A Multi-Document Approach to Generating Diverse Descriptions of Scientific Concepts”!
📄arxiv.org/abs/2205.06982 [1/8] -
Daniel Weld reposted
Harlan Krumholz
@hmkyale
If you are looking for a top review on natural language processing for unstructured data in electronic health records, we have one for you! Led by @YaleLILY_Lab Sr author Drago Radev @YaleSEAS @YaleCompsci @YaleMed @YaleCardiology @comp_science #ai
authors.elsevier.com/c/1foHN5buwWnNn6 -
Daniel Weld reposted
Ai2
@allen_ai
AI2 CEO @etzioni is speaking at @MadronaVentures' first-ever Intelligent Applications Summit on Nov. 2. The Summit will explore topics like emerging trends that matter in #ml & #ai. Request an invite: www.ia40.com/ia40-summit
-
Daniel Weld reposted
Riley Goodside
@goodside
It also works for pirates:
-
Daniel Weld reposted
Riley Goodside
@goodside
Geoffrey Chaucer teaches you sorting algorithms via GPT-3. Only green text is generated.
-
Daniel Weld reposted
Gagan Bansal
@bansalg_
"The striking difference was that developers who used GitHub Copilot completed the task significantly faster–55% faster than the developers who didn’t use GitHub Copilot."
Very promising, real-world results for human-AI interaction!
Eirini Kalliamvakou 🎃
@irina_kAl
Equal parts hard work and exciting work, I'm very glad to be sharing these results! #githubcopilot has had such strong impact on developers on many levels, it's a privilege to have front row seats to how we understand and measure that. More to come!
github.blog/2022-09-07-research-quantifying-github-copilo… -
Daniel Weld reposted
Sahil Bloom
@SahilBloom
What is the single most impactful piece of advice you’ve ever received?
-
Daniel Weld reposted
Tom Goldstein
@tomgoldsteincs
Diffusion models like #dalle and #stablediffusion are state of the art for image generation, yet our understanding of them is in its infancy. This thread introduces the basics of how diffusion models work, how we understand them, and why I think this understanding is broken.🧵
-
Daniel Weld reposted
Zain Kahn
@heykahn
95% of successful storytelling boils down to 7 key elements:
1. A character
2. Their problem
3. Your guide
4. Your plan
5. The call to action
6. Helping them avoid failure and
7. Achieve success
Here's how to craft each element of your story: -
Daniel Weld reposted
Neel Nanda
@NeelNanda5
I've spent the past few months exploring @OpenAI's grokking result through the lens of mechanistic interpretability. I fully reverse engineered the modular addition model, and looked at what it does when training. So what's up with grokking? A 🧵... (1/17) www.alignmentforum.org/posts/N6WM6hs7RQMKDhYjB/a-mechanis…
-
Daniel Weld
@dsweld
Hope to roll this out to full production in @SemanticScholar! Personalized views of authors, conferences, journals, even institutions...
Harmanpreet Kaur
@harmankkaur
Excited to share our #uist2022 paper introducing FeedLens, a novel system that repurposes people’s research paper feeds on Semantic Scholar as lenses. Without any additional training, these lenses show relevance for all types of related entities, like authors and institutions. 🧵
-
Daniel Weld
@dsweld
Great lessons!
Kieran Snyder
@KieranSnyder
I have been managing people off and on for a long time now. Here are 10 open, vulnerable, painful mistakes I’ve made. Ouch. I’m sharing them so you don’t have to make them too. Thread >>
-
Daniel Weld reposted
𝙷𝚒𝚖𝚊 𝙻𝚊𝚔𝚔𝚊𝚛𝚊𝚓𝚞
@hima_lakkaraju
Very excited to introduce, OpenXAI - open-xai.github.io, an open-source framework we have been building for the past year to evaluate/benchmark the faithfulness, stability, fairness of post hoc explanation methods using easy-to-use API & just a few lines of code. [1/N] #xai
-
Daniel Weld reposted
Saleema Amershi
@SaleemaAmershi
Honored to receive this along with Cristina Conati. User modeling for intelligent tutoring systems was where I first learned about all the interesting problems at the intersection of #ai and #hci. Still working on many of these today. Thanks #edm2022!
-
Daniel Weld
@dsweld
Exciting new academic search engine for COVID papers, based on #cord19 icdw.iiit.ac.in/results?size=n_20_n by Kamal Karlapalem's group at IIT Hyderabad
-
Daniel Weld
@dsweld
Nicely written (and insightful!) post on research ideation by @marcotcr medium.com/@marcotcr/coming-up-with-research-ideas-303268…
-
Daniel Weld reposted
David McClure
@clured
I've been playing with the training dataset behind the extremely cool new BLOOM model from @BigscienceW and @huggingface. Here's a sample of 10 million chunks from the English-language corpus, about 1.25% (!!) of the total. Encoded with `all-distilroberta-v1`, then UMAP to 2d.
-
Daniel Weld
@dsweld
Great positions, deadline soon… please apply!
Semantic Scholar
@SemanticScholar
Semantic Scholar Research is looking for HCI / NLP interns and predocs to work on exciting research projects! Topics include accessibility of research papers, tools for literature reviews and info seeking, efficient NLP models, and more! Deadline July 15: allenai.org/careers?team_ai2=semantic+scholar&category_ai…
-
Daniel Weld reposted
Lewis Tunstall
@_lewtun
Excited to share a new tool we’ve built called Evaluation on the Hub 🔥🔥🔥!
With this tool you can evaluate any model on any dataset with any metric🤯
Evaluate your models here👉 huggingface.co/spaces/autoevaluate/model-evaluator
Let’s take a look at how it works 🧵
1/ -
Daniel Weld reposted
Sergey Feldman
@SergeyFeldman
Need a dataset to study academic mentorship? Try S2AMP! The "official" paper is here: dl.acm.org/doi/10.1145/3529372.3533283
Why should you care? Because we have *a lot* of gold data and a huge amount of silver data, and it's all linked to @SemanticScholar.
1/N -
Daniel Weld
@dsweld
+1 to that! Oren @etzioni built an incredible lab with a singular focus on AI for the common good…
Eric Horvitz
@erichorvitz
Fabulous leadership of @allen_ai by Oren @etzioni for nearly a decade. Sad to see his departure. #nlp #ai
-
Daniel Weld reposted
Timothy Snyder
@TimothyDSnyder
Russia has a hunger plan. Vladimir Putin is preparing to starve much of the developing world as the next stage in his war in Europe. 1/16
-
Daniel Weld reposted
manvir singh
@mnvrsngh
In 1966, Marshall Sahlins famously claimed that hunter-gatherers work 15-20 hours a week, which he took as evidence of an “original affluent society”. People (like @SebastianJunger @ThatChrisRyan) still cite this statistic. Yet it’s built on shoddy data & interpretation. [thread]
-
Daniel Weld reposted
Tim Althoff
@timalthoff
Check out @galenweld's research on evaluating causal inference methods for text @icwsm #icwsm2022 at 11:25am ET. We will share an evaluation framework, datasets, results from 648 experiments, and make it all publicly available. See you there!
Galen Weld
@galenweld
Evaluating causal inference methods is challenging, because there is almost never any ground truth. We propose an eval framework at #icwsm2022 with @PeterWestTM, @MariaGlenski, @darbour26, Ryan A. Rossi, and @timalthoff .🎊
arxiv.org/abs/2009.09961
Thread for more ⏬ (🧵1/8) -
Daniel Weld reposted
Timothy Snyder
@TimothyDSnyder
Speaking of the Russo-Ukrainian war, a number of commentators begin from the premise a nuclear power cannot lose a war. This is utter nonsense, and dangerous nonsense. 1/10
-
Daniel Weld reposted
Jason Crawford
@jasoncrawford
The original NYC subway was completed faster than the average environmental impact statement
-
Daniel Weld
@dsweld
After all, Musk works 40h/wk at Tesla, another 40 at SpaceX, another 40 at the Boring Co, and plans at least 40/wk for Twitter...
John Cook
@johnhcook
Elon Musk’s back-to-office edict reignites debate on remote work — here’s what Seattle tech leaders say www.geekwire.com/2022/elon-musks-back-to-office-edict-rei… via @GeekWire
-
Daniel Weld
@dsweld
Don't really understand the details here, but ... I'm going to do it!!
TTI
@TikTokInvestors
What could go wrong?
-
Daniel Weld reposted
Gagan Bansal
@bansalg_
I 100% agree -- its important to understand when explanations help users, for which tasks, and along what metrics.
E.g., it’s important to understand when explanations do and don’t lead to appropriate reliance...
[1/2]
𝙷𝚒𝚖𝚊 𝙻𝚊𝚔𝚔𝚊𝚛𝚊𝚓𝚞
@hima_lakkaraju
Just because one user study showed that explanations produced by a method were not helpful for N homogenous users in a particular context, this does not imply that the method in question has no utility in any other setting. It is important to appreciate this nuance [4/N]
-
Daniel Weld reposted
John Cook
@johnhcook
Amazon sued for ending free Whole Foods delivery as Prime benefit www.geekwire.com/2022/amazon-sued-for-ending-free-whole-f… via @GeekWire
-
Daniel Weld reposted
John Cook
@johnhcook
‘It is not acceptable’: UW computer science program can’t keep up with record demand from undergrads www.geekwire.com/2022/it-is-not-acceptable-uw-computer-sc… via @GeekWire
-
Daniel Weld reposted
Xing Han Lu
@xhluca
Cool feature by @SemanticScholar: Integrated PDF reader where the hyperlinks brings you directly to a paper's semantic page.
-
Daniel Weld
@dsweld
Congratulations Sherrry! Can’t wait to see what you do next!!
Sherry Tongshuang Wu
@tongshuangwu
It's official! I will join @SCSatCMU @cmuhcii as an Assistant Professor in Fall 2022, with a courtesy appointment @LTIatCMU. Toughest decision ever, but I'm excited (albeit scared) to see what comes next :) THANK YOU everyone for all your support!
-
Daniel Weld
@dsweld
I learned so much from Drew, my undergraduate advisor.... will really miss him...
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
I am so very sorry to hear of Drew McDermott's passing. I was literally just thinking of Drew three days back. RIP🙏
Drew McDermott is one of the *sharpest* intellects in #ai. I quote him so very often that many in my group--who have never him--feel as if they did.. 1/ -
Daniel Weld reposted
Sam Altman
@sama
we spent many trillions on the covid response (and suffered huge economic and social damage).
the failure to now spend tens of billions preventing the next pandemic is so depressing. -
Daniel Weld
@dsweld
Very cool application of personalized medicine!
adyn
@adynhealth
We are thrilled to be featured in @ELLEmagazine!
"Personalized birth control is finally available: a new hormonal and genetic test could help predict birth control side effects."
www.elle.com/beauty/health-fitness/a40051559/personalized…
#birthcontrol #sideeffectssuck #reprohealth -
Daniel Weld reposted
Ed Lazowska
@lazowska
UW College of Engineering 2022 Diamond Award Winners Stefan Savage (Distinguished Achievement in Academia) and Justine Sherry (Early Career Award). Congratulations - and thanks for making us look good!
-
Daniel Weld reposted
Huan Sun
@hhsun1
Official now! Excited to be promoted to Associate Professor with tenure @OhioStateCSE @OSUengineering Sincerest thanks to my advisors, mentors, collaborators, and colleagues who wrote support letters for me! A warm thank you to students in @osunlp and to my family @ysu_nlp. ☺️
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
📣 We just launched new Academic Graph datasets, which give access to the full range of data exposed in our API for our entire corpus.
Learn more about these new datasets on the @allen_ai blog:
blog.allenai.org/new-academic-graph-datasets-released-fro… -
Daniel Weld reposted
Amy Zhang
@amyxzh
Really excited for @galenweld to present his work at #icwsm2022 on understanding what redditors value in their community and how those values differ both within and across communities. Key takeaway: no one-size-fits-all!! ➡️
Galen Weld
@galenweld
Looking forward to presenting my work on reddit community values with @amyxzh and @timalthoff next month at #icwsm2022💫
What makes online communities “better”? We surveyed 6k redditors to find out.
arxiv.org/abs/2111.05835 (🧵1/6) -
Daniel Weld
@dsweld
Huge potential here! Hope to incorporate these ideas in @SemanticScholar's production Semantic Reader later this year...
Raymond Fok
@rayrayfok
Can intelligent tools help researchers skim papers?
We developed Scim, an augmented reading interface with faceted highlights that scaffold information discovery during skimming. With @drewmikehead @turingmusician @kylelostat @martihearst & @dsweld. See arxiv.org/abs/2205.04561. -
Daniel Weld
@dsweld
Commentary: Why the courts are right to rule against LinkedIn in battle over data scraping www.geekwire.com/2022/commentary-why-the-court-is-right-t… via @GeekWire
-
Daniel Weld
@dsweld
Love this work!
Tom Hope
@Hoper_Tom
🧐Scientific knowledge keeps expanding, but our mind's capacity remains bounded. In a new paper, we envision #ai systems that assist scientists with information tailored to help with key research tasks. w/ @erichorvitz @dsweld @_DougDowney @etzioni🧵👇
arxiv.org/abs/2205.02007 -
Daniel Weld reposted
@emilymbender.bsky.social
@emilymbender
Actually I said: I applaud the transparency here, not just in releasing the model but also the information about training compute cost and the like. I would hope that the transparency extends to very thorough documentation of the source datasets, as well. www.technologyreview.com/2022/05/03/1051691/meta-ai-large…
-
Daniel Weld reposted
Gagan Bansal
@bansalg_
And its a wrap! We had 80 ppl/8 timezones join in a hybrid setting. Thanks to my co-organizers, speakers, pc, & participants for making this workshop possible @alison_m_smith @ZanaBucinca @tongshuangwu @d19fe8 @JessicaHullman @DrSimoneStumpf
#trait2022 #chi2022 -
Daniel Weld
@dsweld
Really like this paper!
Hyeonsu Kang
@hyeonsu_b_kang
How do scientists decide what papers to read? How does implicit & explicit knowledge of authors & their work factor into those decisions?
We explore these questions in our upcoming #chi2022 paper arxiv.org/abs/2204.10254 @SemanticScholar @allen_ai
1/n -
Daniel Weld reposted
HCI & Design at UW
@uwdub
As we prepare for #chi2022, we're excited to share research from the DUB community!
dub.washington.edu/posts/2022/chi2022papers.html -
Daniel Weld reposted
-
Daniel Weld
@dsweld
Since Microsoft sunset their academic field of study classifier, Semantic Scholar has released an improved version - open source & also available via API blog.allenai.org/announcing-s2fos-an-open-source-academic…
-
Daniel Weld reposted
Amy Zhang
@amyxzh
Academic literature is a form of discourse. Via references, papers situate their work & provide commentary on prior work.
But only 1 direction of discourse is shown when reading a paper. What about all the discussion of the paper from follow-on work?
New work @SemanticScholar ➡️ -
Daniel Weld reposted
Jonathan Bragg
@turingmusician
📢CiteRead, a new paper reading tool @ #iui2022 from @SemanticScholar @allen_ai
Adds margin notes from citing papers written after the paper you're reading
Fun proj w/ @napoldr @amyxzh @dsweld
Q&A w/ Napol @ IUI tmrw 8:30am ET
📄dl.acm.org/doi/10.1145/3490099.3511162
🎥www.youtube.com/watch?v=WjltByXFKAM
Amy Zhang
@amyxzh
Tmrw at #iui2022 @napoldr will present on a new paper reading tool that integrates citation sentences and other info from follow-on papers that cite the current paper. This is work with @turingmusician and @dsweld while Napol interned at @SemanticScholar dl.acm.org/doi/10.1145/3490099.3511162
-
Daniel Weld reposted
Anna Rogers
@annargrs
Dear tech & ethics people: I've done some research on Roskomnadzor: the "ministry of truth" that created the alternative propaganda reality most Russians seem to inhabit. Some truly fascinating stuff. Here's what I learned 🧵
As always, comments/corrections welcome. /1 -
Daniel Weld reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
🆕 Authors now have the option to add their pronouns to their page on Semantic Scholar.
To set this up, first search for and claim your author page, and then visit Settings to add your pronouns.
We're here to help! Reach out with any questions or feedback. -
Daniel Weld reposted
Nils Reimers
@Nils_Reimers
GPT-3 Embeddings by @OpenAI was announced this week.
📈 I was excited and tested them on 20 datasets
😢 Sadly they are worse than open models that are 1000 x smaller
💰 Running @OpenAI models can be a 1 million times more expensive
medium.com/@nils_reimers/openai-gpt-3-text-embeddings-rea… -
Daniel Weld
@dsweld
Nice interview of @_DougDowney on @substack thesequence.substack.com/p/-doug-downeysemantic-scholar-a…
-
Daniel Weld
@dsweld
Excited to see how developer use this new API functionality! Check it out by performing a search at semanticscholar.org
Semantic Scholar
@SemanticScholar
🧵New API release: TLDRs are now available! Plus, three new author fields. Now live:
- TLDRs for 58 million papers in computer science, biology, & medicine. Docs: api.semanticscholar.org/graph/v1
- New author fields: paperCount, citationCount, hIndex. Docs: api.semanticscholar.org/graph/v1 -
Daniel Weld reposted
Maria Gilfoyle
@MariaGilfoyle
Food Investment Thesis 👇🔥🍲💫
Food insecurity costs the health system an additional $53B per year in the US.
Technology can be used to increase access to nutrition while reducing costs for consumers, businesses, and the environment.
changemakerventures.substack.com/p/deep-dive-using-techno… -
Daniel Weld
@dsweld
Congratulations to Scia11y! Converts arbitrary PDF papers to HTML for accessible reading...
sigaccess
@sigaccess
Our judges picked TWO winners for the #assets2021 Best Artifact Award: Adee Accessibility Tool for for Designers (www.adeeapp.com) and Scia11y Paper to HTML Convertor (papertohtml.org).
-
Daniel Weld reposted
Martin Coulter
@MartinJBCoulter
🚨Excl: Behind closed doors, Google’s lawyers are making it a ‘nightmare’ to publish AI research, insiders say.
Sources say the storm over @timnitGebru’s ousting last year has had a chilling effect internally, w/ execs paranoid of another fallout. Read:
www.businessinsider.com/google-ethical-ai-timnit-gebru-20… -
Daniel Weld
@dsweld
I'm thrilled by this new challenge! Come try @SemanticScholar !
Ai2
@allen_ai
We're excited to announce that Dan Weld (@dsweld) is now the General Manager and Chief Scientist for our
@SemanticScholar project! @_DougDowney and @alexwade are joining him to lead research and strategic partnerships – learn more:
medium.com/ai2-blog/new-leadership-for-semantic-scholar-4… -
Daniel Weld
@dsweld
Well, I'm biased of course, but ... great work!
AKBC 2026
@akbc_conf
Congratulations to the authors of the three papers for winning the Outstanding Paper Award at AKBC 2021! #akbc2021
The papers are:
(1/4) -
Daniel Weld reposted
Arman Cohan
@armancohan
Our team @allen_ai is hiring! Consider applying if you are interested in #ai research on scientific literature and beyond. Broad interests across #nlproc, #machineleaning, #hci, IR and CV.
Kyle Lo
@kylelostat
@allen_ai is hiring #nlproc #ml #ai #hci researchers to join @SemanticScholar! We welcome postdocs, scientists, engineers, research interns & predoctoral researchers. Happy to chat if you're interested!
Our team: research.semanticscholar.org
Apply by *Oct 15*: allenai.org/careers -
Daniel Weld
@dsweld
Great study! Reminds me of Mark Neuman's study of ICLR medium.com/@markn_67491
Neil Lawrence
@lawrennd
"Inconsistency in Conference Peer Review: Revisiting the 2014 NeurIPS Experiment"
We logged the publication/citation status of over 1000 papers that were originally submitted to the 2014 conference.
Write up on ArXiv here: arxiv.org/abs/2109.09774 -
Daniel Weld
@dsweld
Incredible challenges & a great community of researchers - non-profit mission: AI for the common good! Come join us!
Kyle Lo
@kylelostat
@allen_ai is hiring #nlproc #ml #ai #hci researchers to join @SemanticScholar! We welcome postdocs, scientists, engineers, research interns & predoctoral researchers. Happy to chat if you're interested!
Our team: research.semanticscholar.org
Apply by *Oct 15*: allenai.org/careers -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
➡️ Check out papertohtml.org, a new on-demand tool from our research team that converts scientific papers from PDF to HTML to increase accessibility.
Share your feedback with our team here: docs.google.com/forms/d/e/1FAIpQLScmhcb-Z8ZAulz4OMGFdvTKy…
#a11y #openscienceweek
Lucy Lu Wang
@lucyluwang
Announcing papertohtml.org, an on-demand PDF to HTML converter for scientific papers! HTML is easier to make accessible for screen readers or on mobile. Let us know what you think! #a11y
Team: @danielking36 @MattLat AlexBuraczynski @codeviking @SemanticScholar @allen_ai -
Daniel Weld reposted
Lucy Lu Wang
@lucyluwang
Announcing papertohtml.org, an on-demand PDF to HTML converter for scientific papers! HTML is easier to make accessible for screen readers or on mobile. Let us know what you think! #a11y
Team: @danielking36 @MattLat AlexBuraczynski @codeviking @SemanticScholar @allen_ai -
Daniel Weld
@dsweld
Machine learning amplifies bioscience; very cool!
Madrona
@MadronaVentures
Congrats to @AAlphaBio on their Series A. We are excited to be working with you! Curious what an Intersection of Innovation company does? - read more here from @mattmcilwain @CPicardo www.madrona.com/a-alpha-bio-and-the-intersections-of-inno…
-
Daniel Weld
@dsweld
Exciting new AI tool aiding human scientific discovery!
Tom Hope
@Hoper_Tom
Keeping track of challenges and directions to solve them is a core part of research. Our new search engine allows researchers and medical professionals to find fine-grained difficulties, limitations and hypotheses in #covid19 literature. challenges.apps.allenai.org @allen_ai
-
Daniel Weld reposted
Lucy Lu Wang
@lucyluwang
please reach out if you have any questions!
Kyle Lo
@kylelostat
@allen_ai is hiring #nlproc #ml #ai #hci researchers to join @SemanticScholar! We welcome postdocs, scientists, engineers, research interns & predoctoral researchers. Happy to chat if you're interested!
Our team: research.semanticscholar.org
Apply by *Oct 15*: allenai.org/careers -
Daniel Weld reposted
Tom Hope
@Hoper_Tom
Silos of scientific knowledge hinder discovery of ideas across research communities. Search & recommendation engines reinforce these filter bubbles.
Our new system, Bridger, helps researchers discover *novel AND relevant* authors that lead to *new ideas*. arxiv.org/abs/2108.05669 -
Daniel Weld
@dsweld
The Semantic Scholar open API has powered many cool 3rd party apps, like @ConnectedPapers ... The inclusion of search makes it that much easier to build upon!
Semantic Scholar
@SemanticScholar
📣 The Semantic Scholar API just got a major upgrade! You can now use it to search our corpus of nearly 200M papers.
Learn more over at the AI2 blog: medium.com/ai2-blog/a-new-and-improved-semantic-scholar-a…
And contact us today about your API project! www.semanticscholar.org/product/api
#nlproc #academictwitter @allen_ai -
Daniel Weld reposted
Amy Zhang
@amyxzh
Excited for @cquanze to present our upcoming paper at #cscw2021! He tackles a longstanding problem of how to deal with disagreement when labeling/grading subjective and ambiguous data. Many methods force some kind of arbitrary tie-breaking in a comparison between two items.
Quan Ze Chen
@cquanze
Crowd judgments and uncertainty come hand in hand.
Sometimes things are just hard to judge and sometimes people disagree.
We introduce Goldilocks, a new process to capture these sources of uncertainty when working with a continuous scale.
arxiv.org/abs/2108.01799
(1/3) -
Daniel Weld reposted
Quan Ze Chen
@cquanze
Crowd judgments and uncertainty come hand in hand.
Sometimes things are just hard to judge and sometimes people disagree.
We introduce Goldilocks, a new process to capture these sources of uncertainty when working with a continuous scale.
arxiv.org/abs/2108.01799
(1/3) -
Daniel Weld reposted
Jonathan Bragg
@turingmusician
📢New preprint:
*FLEX: Unifying Evaluation for Few-Shot NLP*
- rigorous FSL/meta-learning #nlproc benchmark measuring 4 transfer types
- simple baseline w/out heavy prompt engineering / complex meta-learning methods
w/@armancohan,@kylelostat,@i_beltagy
arxiv.org/abs/2107.07170
1/n -
Daniel Weld reposted
Saleema Amershi
@SaleemaAmershi
It's here! Today we launched the #haxtoolkit, practical tools for creating responsible Human-AI eXperiences. aka.ms/haxtoolkit
This was a labor of love, and sometimes sweat and tears, by many @Microsoft. This is the #hax story. Opinions my own. #ai #ux #responsibleai -
Daniel Weld reposted
Duncan Watts
@duncanjwatts
"It's the people, stupid". Very thoughtful and worth parsing carefully.
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Disambiguating author names in scientific literature is a challenge. S2AND, a new model from our Research team, is now live on semanticscholar.org.
Learn more in this post from @danielking36 and @SergeyFeldman:
medium.com/ai2-blog/s2and-an-improved-author-disambiguati… #machinelearning -
Daniel Weld
@dsweld
Great deck! (Alas, ended too soon; i wanted even more!)
Chinmay Kulkarni https://hci.social/@chinmay
@chinmay________
So sometime ago, @msbernst @amyxzh and @andresmh organized an awesome social computing summer camp. They asked me to share a vision slidedeck; and I realized I hadn't shared it broadly: docs.google.com/presentation/d/1IlR1BhvSszyq4QQu-bNSSLLiv…
Commenting is on; would love to know what you think! -
Daniel Weld reposted
Ai2
@allen_ai
PhD student @patrickpflu from @UniofOxford highlights some fantastic study hacks, including our new "beautiful tool" Semantic Reader by @SemanticScholar – check out their fantastic video (Semantic Reader highlighted at 2:22). Thank you, Patrick! 😊
www.youtube.com/watch?v=VQFVcnoo8B0&ab_channel=OxPat -
Daniel Weld
@dsweld
A (happy!) user created a tiktok video for @SemanticScholar's Semantic Reader! www.tiktok.com/@oxpat/video/6962576535551266054
-
Daniel Weld reposted
Anh Totti Nguyen
@anh_ng8
Time to re-evaluate attribution maps and their evaluation metrics! 🙂
Work led by the amazing @GiangNg12491873 , with Daeyoung Kim!
Paper: arxiv.org/abs/2105.14944 -
Daniel Weld reposted
Lucy Lu Wang
@lucyluwang
interested in #scinlp #metascience #textmining #nlproc for scientific text? join us tmr 9-5p ET for the 2nd Scholarly Document Processing Workshop @SDProc @NAACLHLT
keynotes: @yoavgo @IAugenstein @HannaHajishirzi, research track, 3 shared tasks
program: sdproc.org -
Daniel Weld
@dsweld
Not sure how accurate are the sizes, but it looks cool! www.visualcapitalist.com/wp-content/uploads/2021/06/world…
-
Daniel Weld reposted
comp.social
@eegilbert
I took a breath this weekend, and feel like I can reflect on the WL situation. I was concerned about writing this down earlier because I thought it might add fuel to the “Eric and UMSI are out to get me” narrative.
-
Daniel Weld reposted
Harmanpreet Kaur
@harmankkaur
Walter Lasecki is a predator. Read this. Read his "apology" email. If you still don't think that this has been a part of his strategy and manipulation all along, if you can still find it in yourself to give him the benefit of the doubt, then you are a part of the problem.
Geneviève Patterson
@genevievemp
I reported the event in detail to AAAI in 2017 via a video call. Unexpectedly, Walter also emailed me an apology. I sent an email confirming both my interview and his apology to AAAI. I hoped he was contrite. I moved on with my life.
-
Daniel Weld reposted
comp.social
@eegilbert
Important additional context:
- Many new survivors have come forward
- CSE is considering tenuring Lasecki right now
- The survivors who spoke to The Daily are some of the bravest people I know
The Michigan Daily
@michigandaily
“My experience was a part of a pattern of behavior”
The Daily’s first issue of a two part investigation into allegations of sexual misconduct against CSE professor Walter Lasecki.
www.michigandaily.com/news/daily-investigation-finds-dive… -
Daniel Weld reposted
Geneviève Patterson
@genevievemp
Walter Lasecki assaulted me. He pinned me against a bar, he put his hand up my skirt and his fingers in my underwear. He grabbed and twisted. He leaned in and insisted that I sleep with him. “It could be innocent,” he said.
-
Daniel Weld
@dsweld
Inspirational! @zalessya and @WhyLabs rock! www.linkedin.com/pulse/who-said-moms-cant-ceo-alessya-vis…
-
Daniel Weld
@dsweld
Congratulations to @strubell @ananya_g and @andrewmccallum for `Energy and Policy Considerations for Deep Learning in NLP' - inspiring work! api.semanticscholar.org/CorpusID:174802812
Madrona
@MadronaVentures
Madrona has been investing in AI since 2013, today we announced the Inaugural AI Impact Prize for research. Congrats to the winners @ananya__g @strubell @andrewmccallum for uncovering the environmental impacts of AI. www.madrona.com/madrona-inaugural-ai-impact-prize-goes-to…
-
Daniel Weld reposted
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
#semanticreader is based on research from @allen_ai, @UW, and @BerkeleyISchool.
Learn more about the research at 8 am PT today at #chi2021, or watch on YouTube >>
Andrew Head
@drewmikehead
Today at #chi2021, I will be presenting a new project, "Augmenting Scientific Papers with Just-in-Time, Position-Sensitive Definitions of Terms and Symbols". You can catch it at 4pm PST today, 8am tomorrow, or at youtu.be/yYcQf-Yq8B0
Summary follows (1/8) -
Daniel Weld
@dsweld
Congrats to @isabelcachola and the other winners!
-
Daniel Weld
@dsweld
Excited for @danielxli 's news: thedl.danielxli.com/issues/the-dl-special-announcement-fr…
-
Daniel Weld reposted
Ed Lazowska
@lazowska
#allen_ai - Allen Institute for AI takes over Vulcan’s portfolio of environmental big-data projects www.geekwire.com/2021/allen-institute-ai-takes-vulcans-po… via @GeekWire
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Our Research team is involved in community efforts to make #nlp for scientific literature more accessible to researchers and practitioners.
Learn more in this interview with @kylelostat & @lucyluwang ➡️
medium.com/ai2-blog/cord-19-one-year-later-463f8da82cd1 #cord19 -
Daniel Weld
@dsweld
Come join us to build human-centered AI to accelerate research itself!
Kyle Lo
@kylelostat
We're hiring postdocs & research scientists/engineers to join @allen_ai @SemanticScholar! Excited about #nlproc #ml #ai #hci #futureofwork systems that understand scientific literature to assist scholars & accelerate science? Let's chat!
Apply by Mar 30: allenai.org/careers?dept=59084 -
Daniel Weld
@dsweld
Newspaper Navigator was nominated for best `digital humanities dataset’ Please vote for it (third category)
docs.google.com/forms/d/e/1FAIpQLSf9G81BpcPyAbxkM9Jjg63tt… -
Daniel Weld reposted
Chinmay Kulkarni https://hci.social/@chinmay
@chinmay________
I don't usually become a fan of a software system in under 5 minutes, but @SemanticScholar is miles ahead of any similar tool I've used. I love their citation context tool in particular. Try it out!
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
We are hiring researchers and engineers to join the Semantic Scholar team!
Current openings include:
- Engineering Manager
- Software Development Engineer
- Young Investigator
- Research Scientist
allenai.org/careers?dept=59084 -
Daniel Weld
@dsweld
How does political stance REALLY affect behavior wrt factual news?? Read on...
Galen Weld
@galenweld
Excited to share our new work on news sharing behavior on #reddit.
“Political Bias and Factualness in News Sharing Across more than 100,000 Online Communities”
with @MariaGlenski @timalthoff
📃Preprint: arxiv.org/abs/2102.08537
💾Data: behavioral-data.github.io/news_labeling_reddit/
👇Thread: (1/6) -
Daniel Weld reposted
Tim Porter
@tmporter
Huge congrats to the @SaykaraAI team for their acquisition by @NuanceHealth, who together will accelerate the mission of reducing physician burnout through automating clinical documentation with #conversationalai. This combination is a real win-win-win for all parties!
-
Daniel Weld reposted
Daniel Li
@danielxli
You've all read enough about @JeffBezos stepping down as the CEO of @amazon last week, so I won't belabor the topic. Instead, here are my top 10 Jeff Bezos quotes, told in memes.
-
Daniel Weld
@dsweld
Another seminal paper by Jon Kleinberg's group: Dangers of Algorithmic Monoculture: api.semanticscholar.org/CorpusID:231627796
-
Daniel Weld
@dsweld
Proud that @isabelcachola was selected as a finalist for work done at @SemanticScholar !!
-
Daniel Weld reposted
Sherry Tongshuang Wu
@tongshuangwu
So, here’s the preprint I was talking about:
“Polyjuice: Automated, General-purpose Counterfactual Generation”
The title says it all: We develop a “polyjuice” transforming sentences into counterfactuals.
w/ @marcotcr, @jeffrey_heer, @dsweld
homes.cs.washington.edu/~wtshuang/static/papers/2021-arxi…
(1/5) -
Daniel Weld reposted
Alvaro Bedoya
@alvarombedoya
In other words, everyone expected that this would happen. Companies and law enforcement had years to design against it. Yet, here we are. www.nytimes.com/2020/12/29/technology/facial-recognition-…
-
Daniel Weld reposted
Gagan Bansal
@bansalg_
New compelling evidence for developing explainable AI!
Our user-studies on open-domain QA show that explanations help end-users and outperform calibrated confidence (strong, unbeaten baseline) by a significant margin! That too whilst achieving "complementary" performance!! (1/2) -
Daniel Weld
@dsweld
Wrt Google/Timnit controversy, I must agree.
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
Having read @_KarenHao's summary of that parrot paper, I find myself even more in agreement with this 👇take on the whole brouhaha by @yoavgo..
-
Daniel Weld
@dsweld
Nice story on TL;DRs at Nature www.nature.com/articles/d41586-020-03277-2
-
Daniel Weld
@dsweld
Great article by Technology Review on our TL;DR summaries of research papers: www.technologyreview.com/2020/11/18/1012259/ai-summarizes…
-
Daniel Weld
@dsweld
Automatically-generated paper summaries, now on @SemanticScholar! Super helpful, especially on mobile!
Semantic Scholar
@SemanticScholar
Introducing TLDRs on Semantic Scholar, now available in beta for ~10M Computer Science papers!
These auto-generated extreme summaries help you decide which papers are most relevant to your work.
Learn more:
medium.com/ai2-blog/introducing-tldrs-on-semantic-scholar…
#tldr #sciencetwitter -
Daniel Weld reposted
Elisa La Cava
@elisalacava
Incredible profile by @geekwire on our @MadronaVentures founder, Tom Alberg, and his legacy helping build and shape the Seattle tech (and more!) ecosystem www.geekwire.com/2020/tom-alberg-bet-seattle-amazon-shapi…
-
Daniel Weld
@dsweld
"Peer review is an annotation problem" (with no systematic iterative improvement strategy) - insightful perspective on flaws of our review process by @annargrs and @iAugenstein arxiv.org/pdf/2010.03863.pdf
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Considering our Spring/Summer 2021 internship and Pre-doctoral Young Investigator program? The application deadline is November 15. Links to apply:
Research Intern: boards.greenhouse.io/thealleninstitute/jobs/2172007
PYI: boards.greenhouse.io/thealleninstitute/jobs/2171988 -
Daniel Weld
@dsweld
Deadline coming up soon! Please apply!
Jonathan Bragg
@turingmusician
Please RT: Recruiting interns to work on humans+AI problems at @allen_ai with @SemanticScholar researchers like @dsweld, Tom Hope, and myself. Apply by Nov 15 for spring/summer internships in #ai #hci #nlproc boards.greenhouse.io/thealleninstitute/jobs/2172007 (pre/postdoc positions also available)
Semantic Scholar
@SemanticScholar
Apply now for Semantic Scholar's Spring/Summer 2021 internship and Pre-doctoral Young Investigator program. Deadline is November 15, 2020. Links to apply:
Research Intern: boards.greenhouse.io/thealleninstitute/jobs/2172007
PYI: boards.greenhouse.io/thealleninstitute/jobs/2171988 -
Daniel Weld
@dsweld
Very cool blog post (and fun interactive visualization!). Some rejected papers are very, very impactful!
Ai2
@allen_ai
Is there a way to actually quantify the difference between accepted vs rejected #iclr papers?
AI2 researcher and engineer @MarkNeumannnn gives us a glimpse into life after rejection for papers that didn't make the cut, this week on the AI2 Blog:
medium.com/ai2-blog/what-open-data-tells-us-about-rejecte… -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Apply now for Semantic Scholar's Spring/Summer 2021 internship and Pre-doctoral Young Investigator program. Deadline is November 15, 2020. Links to apply:
Research Intern: boards.greenhouse.io/thealleninstitute/jobs/2172007
PYI: boards.greenhouse.io/thealleninstitute/jobs/2171988 -
Daniel Weld reposted
Isabel Cachola
@isabelcachola
Our paper was accepted to #emnlp2020 Findings! The preprint is updated with the camera ready. In the updated version, we've included extensive human eval and additional dataset analysis. We've also collected more data, increasing dataset size to 5.4K TLDRs!
Isabel Cachola
@isabelcachola
I'm exited to announce our paper "TLDR: Extreme Summarization of Scientific Documents"
With @kylelostat @armancohan @dsweld
Paper: arxiv.org/abs/2004.15011
Demo: scitldr.apps.allenai.org
Repo: github.com/allenai/scitldr
(1/4) -
Daniel Weld
@dsweld
Finally, @WhyLabs comes out of stealth mode! Very excited by this company... medium.com/whylabs/introducing-whylabs-5a3b4f37b998
-
Daniel Weld
@dsweld
Maybe i'm biased, but it's a fantastic opportunity - Please apply !
Semantic Scholar
@SemanticScholar
The Semantic Scholar Research team is accepting Spring/Summer 2021 internship applications and 2021 Pre-doctoral Young Investigator applications. The deadline to apply is November 15th, 2020! See our research here: research.semanticscholar.org and apply here: allenai.org/careers?dept=59084
-
Daniel Weld
@dsweld
Fantastic image search tool by @lee_bcg and @LC_Labs !
LC Labs
@LC_Labs
Explore more than 1.5 million newspaper images online! We're so excited to announce the launch of the #newspapernavigator tool, which allows you to find images within millions of historic newspaper pages. Learn more at go.usa.gov/xG58n #chronam #machinelearning @lee_bcg
-
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
A “tell-all” account of improving our academic search engine from @SergeyFeldman on the @allen_ai blog: medium.com/ai2-blog/building-a-better-search-engine-for-s…
-
Daniel Weld
@dsweld
Robot cultural appropriation: www.vogue.com/article/sinead-bovell-model-artificial-inte…
-
Daniel Weld
@dsweld
+100
Thomas G. Dietterich
@tdietterich
Strongly agree with this thread. Especially in the era of online paper presentations
-
Daniel Weld
@dsweld
Interesting...
Ai2
@allen_ai
A "tell-all" account of why improving @SemanticScholar #search is not as simple as you might think...
Dealing with dirty data, feature engineering, proper evaluation, posthoc correction, and more in this article by @SergeyFeldman today on the AI2 Blog:
medium.com/ai2-blog/building-a-better-search-engine-for-s… -
Daniel Weld
@dsweld
Insightful reflections about ML startups www.madronavl.com/stories/anatomy-of-an-ml-startup-part-1 by @jbartot
-
Daniel Weld
@dsweld
A fantastic resource for `whole document' NLP research...
Semantic Scholar
@SemanticScholar
Semantic Scholar researchers curated an open corpus dataset with more than 81M articles across multiple domains for all of your scientific NLP needs. #acl2020nlp #nlproc #textmining #datamining @kylelostat @lucyluwang @MarkNeumannnn Rodney Kinney @dsweld aclweb.org/anthology/2020.acl-main.447
-
Daniel Weld
@dsweld
Disturbing result on XAI. Explanations don't seem to build 'appropriate reliance' - Humans trust the AI more, even when wrong :-(
Gagan Bansal
@bansalg_
New work on explainable AI!
arxiv.org/abs/2006.14779
w/ @tongshuangwu, J. Zhu, R. Fok, @besanushi, @ecekamar, M. Ribeiro, and @dsweld.
When AIs advise people, does an explanation of its reasoning actually help the person? Does it let the human outperform the AI? Does it ...(1/6) -
Daniel Weld
@dsweld
So proud! Great project!!
WA State Library
@WAStateLib
Chronicling America & Navigating Newspapers - Thru multiple Natl Digital Newspaper Program grants from the @NEHgov in conjunction with the @librarycongress, we have contributed over 300,000 pages of digitized WA newspapers to Chronicling America!
blogs.sos.wa.gov/fromourcorner/index.php/2020/06/chronicl… -
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
Is it that disconcerting? Most people publishing in Nature journals are paying a page charge in return for "advertising". At least twitter is free! More to the point, with the information deluge, getting paper suggestions from colleagues you trust is better than from Nature
-
Daniel Weld
@dsweld
Wikipedia is slow to create new pages for concepts like ELMo and Roberta. The first step in automatically generating summary pages is recognizing emerging terms...
Daniel King
@danielking36
New #sigir2020 paper: "High-Precision Extraction of Emerging Concepts from Scientific Literature"
with @dsweld @_DougDowney @semanticscholar
Paper: s2.link/corpusid:219636085
Code and data: github.com/allenai/forecite -
Daniel Weld
@dsweld
Zuck continues his unblemished streak of reprehensible technical innovations...
Rachael Tatman @rctatman@mastodon.rctatman.com
@rctatman
Someone in NLP chose to build a product with a clear use case of curtailing employees right to organize. They could have chosen not to.
As technologists, it is our job to understand the potential uses of products and minimize harm.
theintercept.com/2020/06/11/facebook-workplace-unionize/ -
Daniel Weld reposted
Pedro Domingos
@pmddomingos
When asked why the Allies had won WWII, a Churchill aide replied “because our German scientists were better than theirs”. Let’s not forget that. www.nytimes.com/2020/06/09/technology/china-ai-research-e…
-
Daniel Weld
@dsweld
Exciting search & visualization platform for Covid research papers: www.biorxiv.org/content/10.1101/2020.05.23.112284v2
-
Daniel Weld
@dsweld
Fantastic lineup of speakers for #scinlp - you should submit an abstract!
Kyle Lo
@kylelostat
Interested in #nlproc #datamining #machinelearning #ai over scientific text? We invite you present your work at #scinlp workshop at #akbc2020! Please submit short (1 page) abstracts by June 1st. More details at scinlp-workshop.github.io. Excited about our invited speakers 1/n
-
Daniel Weld
@dsweld
So proud of what @lee_bcg has built - very cool asset for quantitative historians!
TechCrunch
@TechCrunch
Millions of historic newspaper images get the machine learning treatment at the Library of Congress tcrn.ch/2Wd9BYP | by Devin Coldewey
-
Daniel Weld
@dsweld
I really powerful historical resource! I'm excited to see what the hackathon produces on Thursday...
Ben Lee
@lee_bcg
1/ With @LC_Labs, #ndnp, and @dsweld, I'm excited to announce the Newspaper Navigator dataset: extracted visual content & headlines from 16+ million historic newspaper pages in Chronicling America!
Paper: arxiv.org/abs/2005.01583
Code: github.com/LibraryOfCongress/newspaper-navigator -
Daniel Weld
@dsweld
TL;DR - a less accurate AI may make a better teammate! Presents a better loss function for training ML.
Gagan Bansal
@bansalg_
Excited to share a draft of our new work on human-centered AI!
arxiv.org/pdf/2004.13102.pdf
w/ @besanushi @ecekamar @erichorvitz @dsweld
When an AI assists human decision-makers, e.g, by recommending its predictions, is the most accurate AI necessarily the best team-mate? (1/5) -
Daniel Weld
@dsweld
So proud of the @SemanticScholar team for pulling this off! So much opportunity for AI to advance science - especially with this crisis!
Michael Kratsios
@USCTO45
President @realDonaldTrump is unleashing America’s unmatched innovative talent. We are:
☑️Using AI to analyze over 50,000 scholarly articles on coronavirus for insights, including potential therapies
☑️Harnessing American supercomputers to research treatments, cures, & vaccines -
Daniel Weld
@dsweld
Powerful new document-level model! Used in several parts of @SemanticScholar including to make paper recommendations arxiv.org/abs/2003.04315
Arman Cohan
@armancohan
New #acl2020nlp paper: We show that pretraining Transformers on inter-document relatedness signals results in improved doc-level representation learning.
paper: arxiv.org/pdf/2004.07180.pdf
w/ @SergeyFeldman @i_beltagy Doug Downey @dsweld -
Daniel Weld
@dsweld
Very cool work!!
Arman Cohan
@armancohan
Longformer: our new Transformer for long docs replacing full quadratic self-attention with a linear self-attention pattern. Pretraining analysis on long doc nlp tasks, multiple sota results in char lm and qa. w/ @i_beltagy @mattthemathman
arxiv.org/pdf/2004.05150.pdf
summary below: -
Daniel Weld
@dsweld
Exciting visualization tool for exploring the CORD-19 dataset of Coronavirus research medium.com/ai2-blog/coviz-helping-scientists-visualize-an… from @SemanticScholar
-
Daniel Weld
@dsweld
Thoughtful analysis of COVID response strategies
Trevor Bedford
@trvrb
I've been mulling over the @MRC_Outbreak modeling report on #covid19 mitigation and suppression strategies since it was posted on March 16. Although mitigation through social distancing may not solve things I believe we can bring this epidemic under control. 1/19
-
Daniel Weld
@dsweld
Proud of the @SemanticScholar team for rolling out this important resource so quickly!
Ai2
@allen_ai
AI has a critical role to play in fighting COVID-19 - @SemanticScholar has taken the lead in partnership with several groups to produce the COVID-19 Open Research Dataset (CORD-19), a free resource of 29K+ scholarly articles for the #ai research community: pages.semanticscholar.org/coronavirus-research
-
Daniel Weld
@dsweld
Explaining an arbitrary blackbox model is one thing; tuning it is another!
Ben Lee
@lee_bcg
I'm very happy to share our new paper on LIMEADE with my advisor @dsweld as well as @kylelostat and Doug Downey in collaboration @SemanticScholar @allen_ai. What is LIMEADE? arxiv.org/abs/2003.04315
-
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
When a review process sets itself up as a gate keeper and then makes arbitrary decisions, it loses the respect of the community. It also devalues the papers that are accepted @IJCAI
-
Daniel Weld reposted
Practical AI 🤖
@PracticalAIShow
🚢 New episode of Practical AI!
🎙 AI-powered scientific exploration and discovery with @SemanticScholar from @allen_ai
⚡️ featuring @daraymond @chrisbenson @dwhitena
💭 tagged #ai #machinelearning #nlp
💚 changelog.com/practicalai/76 -
Daniel Weld
@dsweld
Nice blog post on the MSR/Microsoft perspective on Human-AI Interaction (it was fun helping with the research & contributing to the tutorial)
www.microsoft.com/en-us/research/project/guidelines-for-h… -
Daniel Weld reposted
Semantic Scholar
@SemanticScholar
Semantic Scholar is getting ready for 2020 and dreaming up great ideas for new features. What do you hope to see from Semantic Scholar in 2020? 🌱💭
-
Daniel Weld
@dsweld
So glad someone is making the point that ML can be LESS biased than people...
𝐒𝐞𝐧𝐝𝐡𝐢𝐥 𝐌𝐮𝐥𝐥𝐚𝐢𝐧𝐚𝐭𝐡𝐚𝐧
@m_sendhil
We used fictitious resumes to find employer bias.
We used statistics to find bias in a healthcare algorithm.
Side by side these studies show why discrimination perpetrated by algorithms is so very different from discrimination perpetrated by people
www.nytimes.com/2019/12/06/business/algorithm-bias-fix.html -
Daniel Weld reposted
Madrona
@MadronaVentures
Geek of the Week features venture partner + Founder of @octoml + @uwcse prof @luisceze on what makes him tick - and his final words illustrate his approach to life! (bottom of the article!) www.geekwire.com/2019/luis-ceze/ @geekwire
-
Daniel Weld
@dsweld
Probably my biggest frustration with Microsoft products
-
Daniel Weld
@dsweld
Nails it.
Yann LeCun
@ylecun
GM is the root cause of the purported war between connectionists and symbolists
He created it to make himself relevant.
But it does not actually exist.
Clamoring that DL should be hybridized with symbol manipulation is useless.
Showing how to do it would be useful. -
Daniel Weld
@dsweld
Very cool & transformational tech!
Haym Hirsh
@haymhirsh
Nice article about the "virtual production" filmmaking used by ILM in The Mandalorian. apple.news/ATxPFlw3HRfS8Vk9nbf2HuA
-
Daniel Weld
@dsweld
Please RT: Apply for a research internship with the
@allen_ai
@SemanticScholar team in Seattle! Multiple openings for topics including
#nlproc,
#humanaiinteraction,
#hci &
#futureofwork for scholars. Apply now (we review applications on a rolling basis):
allenai.org/jobs/job/research-internships-209513.html -
Daniel Weld reposted
Ed Lazowska
@lazowska
Super #geekwire article on Tim Althoff's keynote at #uwallen's Research Showcase on Wedneday - How data from smartphones and apps can inform us about human health on a global scale www.geekwire.com/2019/data-smartphones-apps-can-inform-us… via @GeekWire
-
Daniel Weld
@dsweld
Amy is simply fantastic! So excited that she has joined the UW Allen School....
-
Daniel Weld
@dsweld
I like the way he thinks...
Scott Galloway
@profgalloway
The first trillionaire will figure out how to extend or enhance time.
Grocery shopping takes 69 min/wk. You grocery shop 1.6x/wk, and the average commute for grocery shopping is 12.5 min each way.
Walmart Unlimited Delivery gave us 4 days/yr back.
www.businessinsider.com/nyu-professor-scott-galloway-bill… -
Daniel Weld
@dsweld
Exciting postdoc in NLP/IR/HCI with @MartiHearst at UC Berkeley, collaborating with us at @SemanticScholar on a new paper-reading experience! Apply soon!
aprecruit.berkeley.edu/JPF02405 -
Daniel Weld
@dsweld
Great advice! I had internalized many of these, but there were key points I'd missed...
Adam Marcus
@marcua
I never had a 1:1 before I became a manager and started 1:1s with my mentees. In the intervening years, I learned a few things, and created an N=1 guide here: blog.b12.io/how-to-have-a-1-1-8d52734434b9
Hopefully it shortens the learning curve for new managers a bit. Send feedback! -
Daniel Weld
@dsweld
Exciting (belated) birthday - i'm so happy to be now part of the team!
Semantic Scholar
@SemanticScholar
Semantic Scholar started in 2015 as a research project developed by @allen_ai. Back then, it was just a small team trying to solve the problem of information overload in computer science academic literature... www.washingtonpost.com/news/to-your-health/wp/2015/11/02/…
-
Daniel Weld
@dsweld
Proud!
Ed Lazowska
@lazowska
#uwallen at ACM SIGACCESS 2019 ASSETS Conference: Best Student Paper Award, Impact Paper ("Test of Time") Award, and Best Paper Award - news.cs.washington.edu/2019/11/06/allen-school-accessibil…
-
Daniel Weld reposted
Madrona
@MadronaVentures
Today @theinformation working with @tribecap released their first-ever "Ten Seed Investors to Watch" list and we are pleased to be recognized alongside our colleagues from @fcollective, @Accel, @sequoia and others! www.theinformation.com/articles/the-informations-seed-inv…
-
Daniel Weld
@dsweld
Very proud of @galenweld who just completed climbing the 100 highest mountains in Washington !! news.cs.washington.edu/2019/09/23/peak-performance/
-
Daniel Weld
@dsweld
Scathing, funny and all too true... WeWTF:
-
Daniel Weld
@dsweld
An easy introduction to modern error analysis techniques for ML
Sherry Tongshuang Wu
@tongshuangwu
Check out our new blog post on our #acl2019nlp paper! A ~10min story on how to correctly debug your ML systems, with the help of our Errudite tool. Joint work with @marcotc, @jeffrey_heer and @dsweld; Posted on the @uwdata blog.
link.medium.com/ixO9n2Ua6Y -
Daniel Weld
@dsweld
Debugging your ML system is hard! Check out this powerful tool that can help... medium.com/@uwdata/errudite-55d5fbf3232e
-
Daniel Weld
@dsweld
A little historical context to better understand AI progress...
Rodney Brooks
@rodneyabrooks
The point is that It happened in 1962. James Slagle’s PhD thesis system (under Minsky) beat MIT freshman at the calculus exam. In 1962. So no arguments about “look how well we do at this narrow part, the rest will come soon” convince me in any way.
Thomas G. Dietterich
@tdietterich
Doing a deep dive in any one area of expertise tells us hardly anything about "across-the-board" intelligence. Hence, it will take the same number of years as it would take if we didn't have a system for freshman calculus (i.e., we have no idea how many).
-
Daniel Weld reposted
Tim Althoff
@timalthoff
Incredibly honored to receive the #kdd2019 dissertation award for. Grateful to my dissertation committee @jure, @erichorvitz, @jurafsky, Scott Delp @MobilizeCenter, and @trevorhastie, and @HastieTrevor.
-
Daniel Weld
@dsweld
Continuous span masking and a new learning objective make a huge difference!
Mandar Joshi
@mandarjoshi_
SpanBERT: a new pre-training method for better span representation!
(w/ @danqi_chen @yinhanl @dsweld @LukeZettlemoyer @omerlevy_)
Big gains on QA, SoTA on Coref and TACRED. Better pre-training tasks and objectives without any extra data/params.
arxiv.org/abs/1907.10529 (1/6) -
Daniel Weld reposted
Tom M Mitchell
@tommmitchell
Sad to hear of the passing today of AI pioneer Patrick Winston. His 1970 PhD on machine learning was one of my greatest inspirations when I was a PhD student - he went on to do so much more, including leading MIT's AI lab for 25 years. RIP Patrick.
-
Daniel Weld
@dsweld
Interactive map showing where in the US government is using facial recognition: www.banfacialrecognition.com/map/ what do you think? (some users seem ok to me )
-
Daniel Weld
@dsweld
Very cool company (and excellent NLP/IE technology)!
Madrona
@MadronaVentures
Excited to announce our investment in a new company out of @allen_ai and backed by @wilsonsonsini and Madrona, @LexionAi @tmporter @elisalacava Here's our story www.madrona.com/lexion-funding/
-
Daniel Weld reposted
James Riney🐠Coral Capital
@james_riney
Japanese car rental companies couldn't figure out why their cars were be rented, but not driven.
So they did a survey to find out:
😴naps
👨💻work
📦storage
📱charge phones
🍱eat
🤙talk on the phone
📺watch TV
🎃dress up for halloween
🎶practice rapping
www.asahi.com/ajw/articles/AJ201907040011.html -
Daniel Weld
@dsweld
Many replies here already, but I'll add one observation: arXiv's user survey (arxiv.org/pdf/1607.08212.pdf) shows that the majority of arXiv users _want_ the platform to support this ability to easily find citing papers & blogs.
Zachary Lipton
@zacharylipton
It's fine for Google Scholar and Semantic Scholar and ResearchGate to do this, but I think arXiv should should stay above the fray and out of the curation game, and 1 million miles away from the bibliometrics game.
-
Daniel Weld
@dsweld
Great collection of NLP papers
Victor Sanh
@VictorSanh
A few weeks ago, a friend of mine asked me "Which papers can I read to catch up with the latest trends in modern NLP?". 🏃♂️👨🎓
I compiled a list of papers and resources for him 📚 and thought it would be great to share it!
medium.com/huggingface/the-best-and-most-current-of-moder… -
Daniel Weld
@dsweld
Wow!
-
Daniel Weld
@dsweld
Nice summary of @kgajos and my work on SUPPLE... news.cs.washington.edu/2019/04/15/allen-school-researcher…
-
Daniel Weld
@dsweld
Tom's done an amazing job advising Amazon; amazing to think of how they've changed over 23 years!
Madrona
@MadronaVentures
From the earliest days, through ups and downs, new products and services, founding director Tom Alberg believed in @amazon and served tirelessly on the board for 23 years. He is stepping down this spring to make room for new members. Congrats Tom! www.geekwire.com/2019/early-amazon-investor-tom-alberg-le…
-
Daniel Weld reposted
Haym Hirsh
@haymhirsh
Google's new AI ethics council facing difficulties, most recently with Alessandro Acquisti resigning this weekend. www.latimes.com/business/la-fi-tn-google-artificial-intel…
-
Daniel Weld
@dsweld
Does ML simply need more data/computation? Or more human engineering? Insightful discussion...
Rodney Brooks
@rodneyabrooks
Just posted "A Better Lesson", a short review of Ruch Sutton's recent post "The Bitter Lesson". rodneybrooks.com/a-better-lesson/
-
Daniel Weld
@dsweld
If the GPT2 PR didn’t convince you that the OpenAI sold out, this surely does.
-
Daniel Weld
@dsweld
Akin to the challenges introduced by cameras: “the photography of imaginary things “
Luis Ceze
@luisceze
Super well articulated piece on authorship credit in the age of AI art.
Aaron Hertzmann
@AaronHertzmann
Here is a piece I've written on the challenges of AI art authorship in the age of @_joelsimon's #ganbreeder, and a proposal for "Open Source Art."
theconversation.com/new-ai-art-has-artists-collaborators-… -
Daniel Weld
@dsweld
Insightful post by @scottjacobson on the digital/physical intersection & opportunities in Omni-channel retail: www.madrona.com/investment-themes-future-of-retail/
-
Daniel Weld
@dsweld
Understanding Facebook’s latest ‘privacy first’ proposal: Zuckerberg’s So-Called Shift Toward Privacy www.nytimes.com/2019/03/07/opinion/zuckerberg-privacy-fac…
-
Daniel Weld
@dsweld
How should AI systems interact with people? Read these guidelines!
Microsoft XC Research
@MicrosoftRI
#microsoft’s new Human-AI Interaction Guidelines are more than just a checklist. Learn how to use them creatively! msft.social/zxnIgj #ai #chi2019
-
Daniel Weld
@dsweld
Thought provoking, but will it happen?
comp.social
@eegilbert
I wrote a new short essay: "Just ban the nazis," and the exasperation of sociotechnical platform politics. eegilbert.org/posts/just-ban-the-nazis/
-
Daniel Weld reposted
christine betts
@bettsoff
the first version of my first paper is up on arxiv!
arxiv.org/abs/1902.05170 -
Daniel Weld
@dsweld
A fantastic job!!
Ai2
@allen_ai
Congratulations to AI2's Daniel King, @waleed_ammar, @ssgrn, @bettsoff, @i_beltagy, and Madeleine van Zuylen from our #semanticscholar and #allennlp teams on winning the Rich Text Competition from the Coleridge Initiative! Learn about the competition here: coleridgeinitiative.org/richcontextcompetition
-
Daniel Weld
@dsweld
Agree; great post! And unlike many management books/posts, it was succinct!
-
Daniel Weld
@dsweld
Show me the beef!
-
Daniel Weld
@dsweld
Really excited by our work distilling "Guidelines for human-AI Interaction." CHI Paper = www.microsoft.com/en-us/research/publication/guidelines-f… ;& nice blog post = www.microsoft.com/en-us/research/blog/guidelines-for-huma…
-
Daniel Weld
@dsweld
Beyond deep fakes... RL to generate phishing attacks...via @NYTimes www.nytimes.com/2018/11/19/science/artificial-intelligenc…
-
Daniel Weld reposted
Оlga Belogolova 🌻
@olgs7
“Russian propaganda is cacophonous...Americans assume that propaganda serves a clear, actionable objective...In fact, the purpose of totalitarian propaganda is to take away your ability to perceive reality.” - @mashagessen
-
Daniel Weld reposted
Rowan Zellers
@rown
Today we're announcing Visual Commonsense Reasoning, a new task and large-scale dataset, with the goal to bring AI systems from recognition- to cognition-level understanding. Joint work with @ybisk, Ali Farhadi, and @YejinChoinka.
visualcommonsense.com
arxiv.org/abs/1811.10830 -
Daniel Weld
@dsweld
Super excited to join the Semantic Scholar team at @allen_ai ! www.geekwire.com/2018/leading-university-washington-profe…
-
Daniel Weld
@dsweld
Excited to hear TTI's David McAllester talk about predictive coding and NLP semantics...
-
Daniel Weld reposted
Allen School
@uwcse
#mondaymotivation: Come work with us! #uwallen seeks candidates for multiple tenure-track, tenure-eligible & lecturer positions. Educate the next generation of innovators, do cool research, and be part of our dynamic community: www.cs.washington.edu/faculty_candidates
-
Daniel Weld reposted
Johan Ugander
@jugander
A result from last month that I'd missed: linear programs brought down from n^2.5 to n^omega (matrix multiplication), breaking a barrier that has stood since 1989 (29 years). arxiv.org/abs/1810.07896
-
Daniel Weld
@dsweld
Don’t isolate crowd workers, hoping for high accuracy thru majority vote or EM... Instead, match workers who disagree, using focused, contextual multi-turn argumentation!
Quan Ze Chen
@cquanze
New paper! Cicero: A new workflow that improves crowd accuracy by engaging workers in multi-turn, contextual discussions. Cicero gets significant worker accuracy improvement over existing argumentation systems and also works on multi-choice questions. arxiv.org/abs/1810.10733
-
Daniel Weld
@dsweld
Really interesting!
Wanilla
@WanillaCo
"I’m a Child of Immigrants. And I Have a Plan to Fix Immigration. " by SONIA NAZARIO via NYT www.nytimes.com/2018/10/26/opinion/caravan-migrants-asylu…
-
Daniel Weld reposted
Mandar Joshi
@mandarjoshi_
Why embed word *pairs*? Because QA & NLI often involve reasoning over implied relations (e.g. lexical or common sense) between pairs, and relying solely on end-task data for this background knowledge often results in poor generalization. (2/n)
-
Daniel Weld reposted
Mandar Joshi
@mandarjoshi_
New paper! pair2vec: We train word pair embeddings from predictive contexts, and use them to inject background knowledge into QA & NLI models. We get significant gains on top of ELMo, including SotA on the adversarial Glockner NLI dataset (93.4). arxiv.org/abs/1810.08854 (1/n)
-
Daniel Weld
@dsweld
No surprise here... human-AI interaction is governed by the principles of HCI...
-
Daniel Weld reposted
John Horton
@johnjhorton
Congratulations to @Upwork on their IPO! It's not only a fantastic business built by great people---it also creates economic opportunities around the 🌎, making it a powerful force for overcoming the unfairness of geography & birth. It's an inspiration. techcrunch.com/2018/10/03/upwork-pops-more-than-50-percen…
-
Daniel Weld
@dsweld
Great to be a part of @MadronaVentures ! www.seattlebusinessmag.com/technology/madrona-venture-gro… @seattlebusiness
-
Daniel Weld reposted
Johan Ugander
@jugander
Perhaps I've been spending too much time with mechanism design colleagues, but It sounds to me like registration for industry attendees was way, way underpriced...
NeurIPS Conference
@NeurIPSConf
#nips2018 The main conference sold out in 11 minutes 38 seconds
-
Daniel Weld reposted
Caitlin Hudon
@beeonaposy
New blog entry!💫 In no particular order, here are the coolest things I learned at #jupytercon:
caitlinhudon.com/2018/08/27/coolest-things-jupytercon/ #dslearnings -
Daniel Weld
@dsweld
And I thought Amazon Go was leading the way in retail... Think again!
Rishi Taparia
@taps
1/ I just toured the @AlibabaGroup #newretail grocery concept, Hema, and I still can't pick my jaw up off the floor
-
Daniel Weld
@dsweld
Great insights, clearly written :
Madrona
@MadronaVentures
Interested in connecting directly with your customers and building those relationships? Look to your data. How you use and protect it - check out this piece by @SSomasegar @danielxli venturebeat.com/2018/08/19/how-to-create-a-virtuous-cycle… @VentureBeat
-
Daniel Weld
@dsweld
Fantastic!
Tim Hwang
@timhwang
In collaboration with @knightfdn, the AI Initiative is launching a $750,000 open call for proposals on ensuring that the use of artificial intelligence in the news ecosystem be done ethically and in the public interest
Applications open in Sept! Details:
aiethicsinitiative.org/news/2018/8/5/ai-initiative-announ… -
Daniel Weld reposted
Ed Lazowska
@lazowska
Parking Has Eaten American Cities www.citylab.com/transportation/2018/07/parking-has-eaten-…
-
Daniel Weld reposted
Allen School News
@csenews
New post: Allen School strengthens its leadership in AI with the arrival of Hannaneh Hajishirzi news.cs.washington.edu/2018/07/09/allen-school-strengthen…
-
Daniel Weld
@dsweld
Why hire a celebrity to endorse your product, when you can get an AI to do so? The AI may even have more followers... money.cnn.com/2018/06/25/technology/lil-miquela-social-me… (Is this ok?)
-
Daniel Weld
@dsweld
I agree that Human-AI teams are crucial, but this report misses the key issues for making the tech work: making AI Intelligible. See eg www.semanticscholar.org/paper/Intelligible-Artificial-Int…
-
Daniel Weld
@dsweld
Interesting question : will improved simulators weaken the value of this moat?
Haym Hirsh
@haymhirsh
Waymo: 7,000,000 miles of self-driving cars arstechnica.com/cars/2018/06/waymo-announces-7-million-mi…
-
Daniel Weld
@dsweld
Autonomous cars: As Uber and Tesla get repeated bad news, Waymo ramps up with 62K more vehicles on the road.
arstechnica.com/cars/2018/06/as-uber-and-tesla-struggle-w… via @zacharylipton -
Daniel Weld
@dsweld
And pro-choice groups messaging women who visit a Planned Parenthood clinic...
Ryan Calo
@rcalo
Tech savvy lawyers using location-based ads to target emergency rooms.
www.npr.org/sections/health-shots/2018/05/25/613127311/di… -
Daniel Weld
@dsweld
Worst possible human-in-loop system: uber autonomous car detects pedestrian obstacle but 1) programmed not to brake & 2) doesn't alert human driver www.recode.net/2018/5/24/17389120/uber-fatal-crash-arizon…
-
Daniel Weld reposted
-
Daniel Weld reposted
Marc
@setmurphy
the number of school shootings that have happened since the year 2000 around the world:
ENGLAND: 0
GREECE: 1
NETHERLANDS: 1
SPAIN: 1
INDIA: 1
ARGENTINA: 1
RUSSIA: 1
CHINA: 3
MEXICO: 4
AUSTRALIA: 5
CANADA: 5
GERMANY: 5
SOUTH AFRICA: 5
USA: 213
#texas -
Daniel Weld reposted
Ben Recht
@beenwrekt
Thought provoking blog post on potential unintended consequences of "fair" machine learning algorithms. With illuminating visualizations. bair.berkeley.edu/blog/2018/05/17/delayed-impact/
-
Daniel Weld reposted
Jeff Dean
@JeffDean
Nice analysis by Dario Amodei and Danny Hernandez of the computation scale for a variety of recent important ML papers, and on the trend in amount of compute for these large experiments (doubling every 3 months!).
Oriol Vinyals
@OriolVinyalsML
Very cool analysis on compute vs year for some of the most notable advances in AI. blog.openai.com/ai-and-compute/
-
Daniel Weld reposted
Haym Hirsh
@haymhirsh
Questions over the authenticity of Google's voice demo. apple.news/AU1lqZaRFR12PGQTOMbXx5g
-
Daniel Weld
@dsweld
2/2 Overall the #govertical event was fantastic with 8 strong pitches. Very impressed with 2nd and 3rd place teams Firewise & Healthshop. Excited to see what they do next! www.goverticalworkshop.com/results-2018-ml
-
Daniel Weld
@dsweld
1/2 Congratulations to Hyper AI - winner of the @MadronaVL #govertical Ideation Weekend! Exciting ML tech for improving food safety: www.goverticalworkshop.com/results-2018-ml
-
Daniel Weld
@dsweld
Do i have this right? All computer science (systems, quantum, cryptography etc) are now a subset of AI?
Google AI
@GoogleAI
Hmm, have I made a wrong turn? I was looking for @GoogleResearch…
Nope, you're in the right place! We’re unifying all of our research efforts under “Google AI”, which encompasses all the state-of-the-art innovation happening across Google. Learn more at blog.research.google/2018/05/introducing-google-ai.html -
Daniel Weld
@dsweld
Great summary of current trends for Ethereum & scalable blockchain by @danielxli medium.com/@danielxli/5-takeaways-from-edcon-2018-1d937b8…
-
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
Several machine learning researchers have signed a statement regarding the upcoming launch of Nature Machine Intelligence. If you agree, I encourage you to sign this as well. openaccess.engineering.oregonstate.edu
-
Daniel Weld
@dsweld
Great that this important topic is seeing attention!
-
Daniel Weld
@dsweld
#proud Great talk by @galenweld at #chi2018 on eKichabi
-
Daniel Weld reposted
Tom Bennett OBE
@tombennett71
Absolutely fantastic poster outlining cognitive biases. This should be compulsory to study before studying anything else. Or registering on Twitter. Or reading anything. Pls RT
-
Daniel Weld
@dsweld
Inspiring vision for Artificial Intelligence by UC Berkeley Professor Michael Jordan: medium.com/@mijordan3/artificial-intelligence-the-revolut…
-
Daniel Weld reposted
Alvaro Bedoya
@alvarombedoya
1/ As the lead staffer to the Senate privacy subcommittee, I organized several oversight hearings of tech companies like Facebook, Apple, and Google. Some context for folks who don’t normally watch these kinds of hearings.
-
Daniel Weld
@dsweld
Most Twitter bots are benign (good thing too, because they are powerful)
𝐃𝐚𝐯𝐢𝐝 𝐁𝐫𝐢𝐧
@DavidBrin
This link - not brought to you by a Twitter bot
www.vox.com/technology/2018/4/9/17214720/pew-study-bots-g… -
Daniel Weld
@dsweld
Perspective: Why 2017 Was the Best Year in Human History www.nytimes.com/2018/01/06/opinion/sunday/2017-progress-i…
-
Daniel Weld reposted
Duncan Watts
@duncanjwatts
John Paul Stevens: Repeal the Second Amendment www.nytimes.com/2018/03/27/opinion/john-paul-stevens-repe…
-
Daniel Weld
@dsweld
Great summary of the problems getting deep RL to work www.alexirpan.com/2018/02/14/rl-hard.html via @turingmusician
-
Daniel Weld
@dsweld
I'm excited to help organize the GoVertical Machine Learning Startup Creation Weekend on May 4-6! If you are a budding entrepreneur, apply now to be accepted to this workshop (deadline is March 28).
www.goverticalworkshop.com -
Daniel Weld
@dsweld
Not a breach - but much worse.
New York Times Opinion
@nytopinion
Despite Facebook’s claims, everyone involved in the Cambridge Analytica data-siphoning incident did not give his or her “consent” — at least not in any meaningful sense of the word, says @zeynep www.nytimes.com/2018/03/19/opinion/facebook-cambridge-ana…
-
Daniel Weld reposted
Haym Hirsh
@haymhirsh
New study on bias in online education. "We find that instructors are 94% more likely to respond to forum posts... storage.googleapis.com/gse-website-archive/cepa/cepa.stan…
-
Daniel Weld
@dsweld
I guess I’m biased, but still...
Miles Brundage
@Miles_Brundage
"Intelligible Artificial Intelligence," @dsweld and Bansal: arxiv.org/abs/1803.04263
-
Daniel Weld reposted
Jeffrey Sachs
@JeffreyASachs
There Is No Campus Free Speech Crisis: An Unreasonably Long Thread
-
Daniel Weld reposted
Madrona
@MadronaVentures
We are so excited about the @MadronaVL expansion and increasing our contribution to the Seattle ecosystem - a great piece here that lays out future @MadronaVL plans for an accelerator www.geekwire.com/2018/seattles-top-venture-capital-firm-h… @Taylor_Soper
-
Daniel Weld
@dsweld
Note to self: do not attack robots
Thomas G. Dietterich
@tdietterich
-
Daniel Weld
@dsweld
Indeed!
Erik Brynjolfsson
@erikbryn
Drone technology is advancing rapidly.
Prediction: there will some serious unintended consequences
#2ma #ai #drones
www.youtube.com/watch?v=Gh5pAT1o2V8 -
Daniel Weld
@dsweld
Same techniques used to make adversarial images for ML classifiers can be used to hack the human brain!
Evan Ackerman
@BotJunkie
Some mind-bending research by @goodfellow_ian and others @GoogleBrain about how adversarial images that fool CNNs can sometimes fool humans as well @IEEESpectrum spectrum.ieee.org/the-human-os/robotics/artificial-intell…
-
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
We need many different levels of replication and reproducibility in machine learning. This article reports on a recent @AAAI2018 workshop on the topic. www.sciencemag.org/news/2018/02/missing-data-hinder-repli…
1/ -
Daniel Weld reposted
T. K. Rengarajan 🇺🇸
@trengarajan
Moving piece by the talented Christine Yared.
Hearing so powerfully from the kids in Parkland, FL will hopefully precipitate action.
God knows we need to fix this.
The most prosperous country in the world needs to be a beacon of inspiration. www.nytimes.com/2018/02/18/opinion/florida-school-shootin… -
Daniel Weld reposted
Tim Porter
@tmporter
It's very exciting to welcome long-time friend and collaborator @luisceze to the @MadronaVentures team. Luis has helped us look at ML/AI & other deep tech companies for some time, but this makes it official. :-) www.madrona.com/madrona-expands-team-adds-talent-director…
-
Daniel Weld
@dsweld
Living in a smart home = badgered by your coffee maker & sleep habits reported to advertisers :-( gizmodo.com/the-house-that-spied-on-me-1822429852
-
Daniel Weld reposted
Panos Ipeirotis
@ipeirotis
Wow! Smart replies now account for more than 10% of the replies on GMail. #wsdm2018
-
Daniel Weld reposted
Anish Athalye
@anishathalye
Defending against adversarial examples is still an unsolved problem; 7/8 defenses accepted to ICLR three days ago are already broken: github.com/anishathalye/obfuscated-gradients (only the defense from @aleks_madry holds up to its claims: 47% accuracy on CIFAR-10)
-
Daniel Weld
@dsweld
Wow. Starts off a bit predictable a then... Wham!
Nando de Freitas
@NandoDF
There is a price to pay for a rich, full, meaningful life. It’s worth paying. I even marked this down with a tattoo. I highly recommend this talk whose content was transformative for me.
TED Talks
@TEDTalks
"Discomfort is the price of admission to a meaningful life." @SusanDavid_PhD
Say no to forced positivity. Here's why you should embrace your "negative" feelings: t.ted.com/Gh7GT4fThis media couldn't be displayed. See it on the original post. -
Daniel Weld
@dsweld
Great study...
Panos Ipeirotis
@ipeirotis
posted: "How many Mechanical Turk workers are there?" www.behind-the-enemy-lines.com/2018/01/how-many-mechanica…
-
Daniel Weld reposted
Jaime Teevan
@jteevan
Friday afternoon is the perfect time for a break. Here's something to entertain you while you avoid work: www.geekwire.com/2018/geek-week-take-break-read-jaime-tee…
-
Daniel Weld reposted
Haym Hirsh
@haymhirsh
Tech news: ad tracking software can access autofill info like email account information saved in your browser. blog.citp.princeton.edu/2017/12/27/no-boundaries-for-user…
-
Daniel Weld reposted
Rodney Brooks
@rodneyabrooks
My January 1st predictions for the coming 32 years. rodneybrooks.com/my-dated-predictions/
-
Daniel Weld
@dsweld
Eight interesting predictions for 2018. My favorite: cameras increasingly taking pictures for machines rather than people www.theatlantic.com/technology/archive/2017/12/8-overly-c…
-
Daniel Weld reposted
Thomas G. Dietterich
@tdietterich
"Even the most die-hard blockchain enthusiasts actually want a bunch of humans arguing about the underlying intention behind a contract, rather than letting the software self-execute. " hackernoon.com/ten-years-in-nobody-has-come-up-with-a-use…
-
Daniel Weld reposted
Andrew McAfee 🇺🇦
@amcafee
An uncomfortable truth: algorithms that disproportionally flag members of a group might NOT be unfairly biased against that group:
www.nytimes.com/2017/12/20/upshot/algorithms-bail-crimina… -
Daniel Weld
@dsweld
Latest from China: "Amazon's consumer tracking with an Orwellian political twist" www.wired.co.uk/article/chinese-government-social-credit-… via Gagan Bansal
Daniel Weld
@dsweld
It's "gamified obedience" - a method of social control dressed up as a points-reward system.
-
Daniel Weld reposted
Jeff Clune
@jeffclune
RANDOM search beats Deep Reinforcement Learning on some Atari games! www.youtube.com/watch?v=CGHgENV1hII One of the crazier results from our new research eng.uber.com/deep-neuroevolution/
-
Daniel Weld reposted
Ben Recht
@beenwrekt
This Ted Chiang essay is the best AI thinkpiece I’ve read all year. Ethics. Optimization. Capitalism. The Singularity. Incredible. www.buzzfeed.com/tedchiang/the-real-danger-to-civilizatio…
-
Daniel Weld
@dsweld
He’s being pretty hard on the deep-mind folks, IMO, but they are good and interesting observations...
Chris Anderson (in London)
@chr1sa
Some very good points in this skeptical analysis of the AlphaZero chess results by
@CamachoCollados medium.com/@josecamachocollados/is-alphazero-really-a-sci… -
Daniel Weld reposted
Jeff Dean
@JeffDean
Slides from my talk in yesterday's ML Systems workshop are now up at learningsys.org/nips17/assets/slides/dean-nips17.pdf
#nips2017 -
Daniel Weld reposted
-
Daniel Weld reposted
Daniel Lowd
@dlowd
Dolphins achieve super-human swimming ability in < 24 hours! It's just a matter of time before they surpass humans in every task. #dolphinzero
-
Daniel Weld reposted
Panos Ipeirotis
@ipeirotis
From zero to best chess player with just 4 hours of training, and no domain knowledge added except for the rules of the game. Bring the popcorn...
Demis Hassabis
@demishassabis
no opening book, no endgame database, no heuristics, no nothing! full paper coming soon, will have things like early games.
-
Daniel Weld reposted
Eric Horvitz
@erichorvitz
Understanding & addressing biases learned by #ai systems. Lucid article www.bloomberg.com/news/articles/2017-12-04/researchers-co… by @dinabass Care needed to understand how ML can learn subtle and powerful biases in our society. @compcomcon @TheOfficialACM @PartnershipAI @ACLU
-
Daniel Weld reposted
Google AI
@GoogleAI
Should the reviewers of a piece of scientific work be made aware of the identify of the authors? Check out the results of a controlled experiment that compares reviewer behavior under single-blind and double-blind review. research.googleblog.com/2017/11/understanding-bias-in-pee…
-
Daniel Weld reposted
𝐃𝐚𝐯𝐢𝐝 𝐁𝐫𝐢𝐧
@DavidBrin
Mastery comes from trying 10,000 experiments - not just 10,000 hours of practice
medium.com/the-mission/forget-about-the-10-000-hour-rule-… -
Daniel Weld
@dsweld
Advanced ML + social media + sensors = better dating?
gizmodo.com/the-future-of-online-dating-is-unsexy-and-bru… -
Daniel Weld
@dsweld
Terrifying. (but I should have known something like this was in the cards)
Tadayoshi (Yoshi) Kohno
@yoshi_kohno
Using ad targeting for surveillance — or — How Alice can buy ads to track Bob (w/ @franziroesner & @covertvines) adint.cs.washington.edu
-
Daniel Weld reposted
𝐃𝐚𝐯𝐢𝐝 𝐁𝐫𝐢𝐧
@DavidBrin
Well here’s a time sink: an apparently addictive game where you play an #ai to maximize the production of paperclips
www.theverge.com/tldr/2017/10/11/16457742/ai-paperclips-t… -
Daniel Weld reposted
Ryan Calo
@rcalo
AAAI/ACM Conference on AI, Ethics, and Society call for papers www.aies-conference.com
-
Daniel Weld
@dsweld
Great suggestions, also raises many questions! Eg, how can one ensure that turning off an AI doesn't cause more harm than leaving it on?
-
Daniel Weld reposted
erikashaffer
@erikashaffer
A 3,000 mile father daughter bike journey for orca whales is almost done in Boston www.patriotledger.com/news/20170824/father-daughter-duos-… @RideForTheOrcas @BostonGlobe
-
Daniel Weld
@dsweld
Congrats to Andrey Kolobov - NYT features his research on MDP-powered autonomous planes! www.nytimes.com/2017/08/16/technology/microsoft-teaches-a… via @rao2z
-
Daniel Weld reposted
Johan Ugander
@jugander
Advances in predictive analytics fundamentally change the nature of risk-pooled health insurance, by @mathbabedotorg www.bloomberg.com/view/articles/2017-08-04/big-data-is-co…
-
Daniel Weld
@dsweld
Hits a nerve! I'd noticed a similar barrage on the Fox News website (I try to visit daily) overwhelming distracting click bait. Now I know Y
Molly Roberts
@mollyeroberts
our work on online propaganda in China in Vox: www.vox.com/world/2017/8/2/16019562/china-internet-freedo… @kinggary @jenjpan
-
Daniel Weld
@dsweld
Thoughtful spin on what I had disdained (via @andyjko )
danah boyd
@zephoria
I left MIT's Media Lab 15 years ago. I learned an important skill there: "demo-or-die": www.linkedin.com/pulse/how-demo-or-die-helped-my-career-d…
-
Daniel Weld
@dsweld
Scary stuff!
Ian Goodfellow
@goodfellow_ian
Adversarial examples that make stop signs misclassified 100% of the time, from multiple viewpoints: arxiv.org/pdf/1707.08945v1.pdf
-
Daniel Weld reposted
Madrona
@MadronaVentures
Congratulations to the entire @Redfin team!
Nasdaq Exchange
@NasdaqExchange
.@Redfin ringing the @Nasdaq Opening Bell in celebration of their IPO! 🔔🎉🎉 $RDFN #nasdaqlisted
-
Daniel Weld reposted
Justin Wolfers
@JustinWolfers
“Wisconsin is paying as much [to Foxconn] as $1 million per job, which will carry an average salary of $54,000.” www.bloomberg.com/news/articles/2017-07-27/america-and-th…
-
Daniel Weld
@dsweld
Fantastic new resource for research on NLP Q/A!
Mandar Joshi
@mandarjoshi_
Leaderboard now up at Codalab. Submit your predictions to get test set eval! competitions.codalab.org/competitions/17208
This post is unavailable. See it on X. -
Daniel Weld
@dsweld
Scary-cool research that subverts ML
OpenAI
@OpenAI
A crafted image that fools an ML classifier even after being printed out, zoomed, and rotated: blog.openai.com/robust-adversarial-inputs/.
-
Daniel Weld reposted
Johan Ugander
@jugander
On human+machine decision making in e.g. "freestyle chess," by @jasonacollins: behavioralscientist.org/dont-touch-computer/
-
Daniel Weld reposted
Google AI
@GoogleAI
Introducing Facets, an open source visualization tool to help understand and analyze #machinelearning datasets research.google/blog/facets-an-open-source-visualization-…
-
Daniel Weld
@dsweld
Insightful & concrete ideas for those considering starting a company!
Madrona
@MadronaVentures
Have an idea for a business and a weekend? @ryanmetzger has a 3 step process for testing it out. medium.com/startup-grind/a-3-step-process-to-test-your-st… @StartupGrind
-
Daniel Weld reposted
Amy J. Ko 🏳️🌈 🏳️⚧️ 🌮 🐈
@amyjko
The Seattle Times welcomes @aninddey back to the Pacific NW to lead the @UW_iSchool and the broader iSchool movement www.seattletimes.com/education-lab/uw-ischools-new-dean-o…
-
Daniel Weld reposted
Ryan Metzger
@ryanmetzger
This article by @BuzzSumo is a gold mine of info on content promotion.buzzsumo.com/blog/went-viral-lessons-promoting-content-in… Many of the tactics I use personally
-
Daniel Weld reposted
Backchannel
@backchnnl
Wikipedia used to be the one place where the internet could agree on facts. No more. www.wired.com/story/welcome-to-the-wikipedia-of-the-alt-r…
-
Daniel Weld
@dsweld
Facebook's hybrid Artificial Intelligence work flow finds Extremist Posts - nytimes.com mobile.nytimes.com/2017/06/15/technology/facebook-artific…
-
Daniel Weld reposted
Max Roser
@MaxCRoser
"14 charts that should restore your faith in humanity"
nordic.businessinsider.com/charts-restore-faith-humanity-… -
Daniel Weld reposted
Allen School News
@csenews
New post: Allen School privacy researchers shed light on secret surveillance with SeaGlass news.cs.washington.edu/2017/06/02/allen-school-privacy-re…
-
Daniel Weld
@dsweld
And happy that @MadronaVentures funded these winners!
Ed Lazowska
@lazowska
A good day for #uwcse #uwallen graduate program alums Mike Cafarella, Raphael Hoffman, Chris Re, and Frits Habermann news.cs.washington.edu/2017/05/15/may-15-a-good-day-for-a…
-
Daniel Weld reposted
Alex Steffen
@AlexSteffen
This is genuine insight.
Censorship isn't burning down the newspaper, anymore, it's clogging our feeds with untruths. -
Daniel Weld
@dsweld
Thought provoking (and encouraging too!)
John Horton
@johnjhorton
AI, labor, & the parable of the horse john-joseph-horton.com/ai-labor-and-the-parable-of-the-ho…
-
Daniel Weld reposted
Johan Ugander
@jugander
Jennifer Widom's blog from her worldwide teaching tour: medium.com/@profwidom Fascinating read about differences of educational culture
-
Daniel Weld
@dsweld
Very exciting work!
Amy J. Ko 🏳️🌈 🏳️⚧️ 🌮 🐈
@amyjko
Excited to share our #chi2017 work on Genie, which enables the use of interactive websites via *any* input modality dl.acm.org/citation.cfm?id=3025506&CFID=751278171&CFTOKEN…
-
Daniel Weld reposted
Ed Lazowska
@lazowska
#uwcse #uwallen Allen School's @jeffrey_heer Jeffrey Heer wins ACM Grace Murray Hopper Award - news.cs.washington.edu/2017/05/03/allen-schools-jeffrey-h…
-
Daniel Weld reposted
Tom M Mitchell
@tommmitchell
congratulations @erichorvitz on new position as Head of all Microsoft Research outside Asia! qz.com/973005/microsofts-new-head-of-research-has-spent-h…
-
Daniel Weld reposted
sarah guo
@saranormous
1/ An unpopular point of view: many enterprise "AI products" and "machine intelligence" products built today have limited appeal or impact
-
Daniel Weld reposted
Sarah A. Evans, PhD & MLIS
@sarahamber
Great explanation of human behavior & our complicated thinking patterns- Who Are You Calling Anti-Science? blogs.scientificamerican.com/guest-blog/who-are-you-calli… #science
-
Daniel Weld reposted
Tadayoshi (Yoshi) Kohno
@yoshi_kohno
Great talk on online dating + privacy Camille, last week at WWW 2017!! Paper here: homes.cs.washington.edu/~cobbc12/publications/WWW2017_Onl…
-
Daniel Weld reposted
Nate Silver
@NateSilver538
All airlines overbook but some are MUCH better at finding volunteers, paying $$$, and avoiding a @United situation. www.transportation.gov/sites/dot.gov/files/docs/resources…
-
Daniel Weld reposted
Stanford NLP Group
@stanfordnlp
All lectures & materials for CS224N: NLP with Deep Learning by @chrmanning & @RichardSocher: web.stanford.edu/class/cs224n/ www.youtube.com/playlist?list=PL3FW7Lu3i5Jsnh1rnUwq_TcylN…
-
Daniel Weld
@dsweld
Great talk/slides by @chrmanning on NN representations for language! nlp.stanford.edu/~manning/talks/Simons-Institute-Manning-…
-
Daniel Weld reposted
Rover.com
@RoverDotCom
Big news: Rover & DogVacay are joining forces! Learn more about this exciting news: www.rover.com/joining-forces
-
Daniel Weld
@dsweld
Great advice for ML practitioners martin.zinkevich.org/rules_of_ml/rules_of_ml.pdf
-
Daniel Weld reposted
Eric Schultz
@EricSchultz
Not one woman in sight as they propose removing maternity coverage.
Vice President Mike Pence Archived
@VP45
Appreciated joining @POTUS for meeting with the Freedom Caucus again today. This is it. #passthebill
-
Daniel Weld
@dsweld
Robocalls get more sophisticated. www.latimes.com/business/lazarus/la-fi-lazarus-chatbot-ph… @Davidlaz
-
Daniel Weld
@dsweld
Drive Magazine's video interview with @adamweld on the design of Hoverbot.io quadcopter www.thedrive.com/aerial/8461/we-talked-with-adam-weld-co-…
-
Daniel Weld reposted
Ed Lazowska
@lazowska
.@UWCSE honored to become the Paul G. Allen School of Computer Science & Engineering! #thankspaul www.washington.edu/news/2017/03/09/paul-g-allen-school-of…
-
Daniel Weld reposted
Joshua Gans
@joshgans
Google is acquiring data science community Kaggle feedproxy.google.com/~r/Techcrunch/~3/Z323uxXbr4Y/
-
Daniel Weld
@dsweld
Fascinating and insightful : www.nytimes.com/2017/03/03/opinion/sunday/why-we-believe-…
-
Daniel Weld reposted
Ryan Calo
@rcalo
Interview of my @uw colleague @dsweld on beneficial artificial intelligence. futureoflife.org/2017/01/29/dan-weld-interview/
-
Daniel Weld
@dsweld
Great tips for advertising on Facebook by @ryanmetzger (@MadronaVentures) who analyzed 2 years of data! medium.com/madrona-venture-group/5-things-i-learned-from-…
-
Daniel Weld reposted
zeynep tufekci
@zeynep
haha if anyone asks why the "pseudo" part in psedo random number generators matter. How to cheat at slot machines. www.wired.com/2017/02/russians-engineer-brilliant-slot-ma…
-
Daniel Weld reposted
Sam Adams
@SamuelAAdams
It is hard to think of anything more contemptible than a person born into wealth denying poor children the right to a good education.
-
Daniel Weld
@dsweld
Another historic win for AI: strategic reasoning w/ incomplete info, bluffing. Libratus beats pro poker players: www.cmu.edu/news/stories/archives/2017/january/AI-beats-p…
-
Daniel Weld
@dsweld
My father-in-law died peacefully this morning. Sad, but proud too. He was such an inspiration! www.nytimes.com/2017/01/27/science/arthur-rosenfeld-dead-…
-
Daniel Weld
@dsweld
Wrote up a few words / photos about the past year (ummm, past two years...) homes.cs.washington.edu/~weld/photos/RosenWeld2016.pdf
-
Daniel Weld
@dsweld
Tell Donald Trump that Americans -do- care about his tax returns. Please sign the whitehouse.org petition petitions.whitehouse.gov/petition/immediately-release-don…
-
Daniel Weld
@dsweld
Very proud: my son, @adamweld, just launched his startup's first product at hoverbot.io - an ultralight indoor quadcopter drone
-
Daniel Weld
@dsweld
Great news! @jbartot joining @MadronaVL to build #ml applications and spinout great companies: www.geekwire.com/2016/madrona-labs-hires-startup-vet-jay-… @geekwire
-
Daniel Weld
@dsweld
Took me 28 years, but I finally finished climbing Washington's 100 highest mountains!! See my retrospective photos.google.com/share/AF1QipNizgNC6zNXpBwlVDeEOzmpibM0U…
-
Daniel Weld reposted
Johan Ugander
@jugander
From playbook of @sinanaral & @dylanwalker, one can start a fake news site to get external view of such diffusion: science.sciencemag.org/content/337/6092/337.abstract
-
Daniel Weld
@dsweld
Fascinating & disturbing.
zeynep tufekci
@zeynep
2016 was a close election where filter bubbles & algorithmic funneling was weaponized for spreading misinformation. www.niemanlab.org/2016/11/the-forces-that-drove-this-elec…
-
Daniel Weld
@dsweld
Don't wait to find out! Real-time vote estimates from VoteCastr www.slate.com/votecastr_election_day_turnout_tracker.html
-
Daniel Weld
@dsweld
How to design a better reputation system for crowdsourcing humancomputation.com/blog/?p=9263
-
Daniel Weld
@dsweld
Samsung buys personal assistant startup, Viv Labs! www.ft.com/content/8c706af8-8b59-11e6-8cb7-e7ada1d123b1
-
Daniel Weld
@dsweld
Insightful observations on ways to grow your startup company: medium.com/madrona-venture-group/10-lessons-from-1-year-i… by @madrona's @ryanmetzger
-
Daniel Weld reposted
anand kulkarni 🇺🇸
@polybot
Instead of claiming a bug bounty, this security firm simply shorted the company's stock and disclosed to the FDA. www.bloomberg.com/news/articles/2016-08-25/in-an-unorthod…
-
Daniel Weld
@dsweld
Thoughtful post on bot-based interfaces x.com/TechCrunch/statuses/768448047990747136 @danielxli
TechCrunch
@TechCrunch
The new paradigm for human-bot communication tcrn.ch/2bfiVXc by @danielxli
-
Daniel Weld reposted
Jayant Krishnamurthy
@jayantkrish
What questions do you wish semanticscholar.org could answer? Eg, "who has written the most ACL papers?" RTs appreciated.
-
Daniel Weld
@dsweld
Algorithmic affirmative action papers.ssrn.com/sol3/papers.cfm?abstract_id=2795203
-
Daniel Weld reposted
Ryan Calo
@rcalo
Mordor
The Verge
@verge
Visa created a ring that can make purchases www.theverge.com/circuitbreaker/2016/6/2/11846194/visa-pa…
-
Daniel Weld
@dsweld
Now more than ever!
Kontra
@counternotions
1994: "…leveraging the softbot's AI capabilities to develop an expressive, goal oriented…interface to the internet"
homes.cs.washington.edu/~weld/papers/cacm.pdf -
Daniel Weld
@dsweld
A great survey of the growing ecosystem of intelligent apps: techcrunch.com/2016/05/24/the-intelligent-app-ecosystem-i… #ai #ml @ssomasegar @DanilexLi @madronaventures
-
Daniel Weld reposted
Ryan Calo
@rcalo
See ⬆ for livestream of today's @WhiteHouse @TechPolicyLab event and/or follow #ai. 🤖🕹
-
Daniel Weld
@dsweld
My post is live on @geekwire: Why "AI will enslave us" arguments miss the true danger
www.geekwire.com/2016/guest-commentary-real-threat-artifi… -
Daniel Weld
@dsweld
Facebook hires journalists (to train ML automation) gizmodo.com/want-to-know-what-facebook-really-thinks-of-j…
-
Daniel Weld
@dsweld
Congratulations to @kgajos - now tenured at Harvard! seas.harvard.edu/news/needleman-and-gajos-promoted-full-p…
-
Daniel Weld
@dsweld
How to attack AI - Microsoft's chatbot learns racism by interacting with people on the Internet. huff.to/1LJDIOf
-
Daniel Weld
@dsweld
Excited to be advising Voicebox!
Voicebox
@Voicebox_Tech
We're putting together a world-class Advisory Board! #nlu #ai www.voicebox.com/voicebox_news/voicebox-announces-scienti…
-
Daniel Weld reposted
Scott Klemmer
@DesignAtLarge
Crowdsourcing outcome depends a lot on process. Prototype twice, launch once... #nameourship www.nytimes.com/2016/03/22/world/europe/boaty-mcboatface-…
-
Daniel Weld
@dsweld
An app store for conversational AI 'bots - with 800M possible customers! tcrn.ch/1U9xOZK via @mattmcilwain
-
Daniel Weld reposted
UW HAI
@uw_hai
Google selling Boston Dynamics! Goodbye robots :-( www.bloomberg.com/news/articles/2016-03-17/google-is-said… (via Dieter Fox)
-
Daniel Weld
@dsweld
Cohort analysis by @ipeirotis shows decline in # requestors on Amazon MechTurk x.com/ipeirotis/statuses/704330861924229120
Panos Ipeirotis
@ipeirotis
A Cohort Analysis of Amazon Mechanical Turk Requesters: www.behind-the-enemy-lines.com/2016/02/a-cohort-analysis-…
-
Daniel Weld reposted
Ed H. Chi
@edchi
Turmoil at Wikipedia and Wikimedia Foundation. motherboard.vice.com/read/wikimedia-foundation-executive-…
-
Daniel Weld
@dsweld
Great job, Yejin!
Allen School News
@csenews
Post Edited: UW CSE’s Yejin Choi named one of IEEE’s “10 to Watch” in AI news.cs.washington.edu/2016/02/10/uw-cses-yejin-choi-name…
-
Daniel Weld
@dsweld
I kept waiting for the robot to SMACK the person!
UW HAI
@uw_hai
Very impressive new humanoid robot from Boston Dynamics! www.youtube.com/watch?v=rVlhMGQgDkY
-
Daniel Weld
@dsweld
The only thing that can stop bad AI... is more AI on.recode.net/1mdO64r @etzioni
-
Daniel Weld
@dsweld
How AI can improve the world - great talk on Computational Sustainability by Carla Gomes www.youtube.com/watch?list=PLTPQEx-31JXgZhR5LWgp_TGLtxdEU… via @uwcse_ai
-
Daniel Weld
@dsweld
Oh course, I'm biased, but ... it's a great talk!
UW HAI
@uw_hai
We're finally getting UW AI seminar talks up on youtube. See @dsweld on "Decision-Theoretic Crowdsourcing" www.youtube.com/watch?v=RK--7PsX6Uk
-
Daniel Weld reposted
Freakonomics
@Freakonomics
Replacing a mediocre teacher with a great one will increase future earnings of the class by $1.5M. freakonomics.com/2016/01/28/is-americas-education-problem…
-
Daniel Weld reposted
Jon Erlichman
@JonErlichman
Amazon Annual Revenue (in billions)
2015: $107.0B
2014: $88.99B
2013: $74.45B
2012: $61.09B
2011: $48.08B
2010: $34.20B
2009: $24.51B -
Daniel Weld
@dsweld
Amazing achievement for AI! By combining deep NN eval with Monte Carlo tree search, prgm beats Europe's Go Champ!
idp.nature.com/transit?redirect_uri=https%3A%2F%2Fwww.nat… -
Daniel Weld reposted
Carlos Guestrin
@guestrin
Really great article from @pmddomingos on #machinelearning camps. Awesome read.
Pedro Domingos
@pmddomingos
All of machine learning in one page - my Wired article about the race for the master algorithm: www.wired.co.uk/magazine/archive/2016/01/ideas-bank/maste…
-
Daniel Weld reposted
Steven Levy
@StevenLevy
If you care about food--or eat--read (& follow!) @kimbal musk's new @medium publication--Food is the New Internet.medium.com/food-is-the-new-internet/why-food-is-the-new-i…
-
Daniel Weld reposted
MIT CSAIL
@MIT_CSAIL
.@Toyota's $1bn AI team to include fmr CSAIL director @rodneyabrooks: for.tn/1Z9dxXc
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Cool map of #seatechuniverse 2015! Exciting how the region has expanded & diversified. www.bioxconomy.com/seattle/2015/12/02/new-map-shows-expan… . @WTIA @MadronaVentures
-
Daniel Weld reposted
anand kulkarni 🇺🇸
@polybot
Themes from #hcomp2015: crowds should work collaboratively; crowds are building new systems for us (vs doing all the work themselves)
-
Daniel Weld
@dsweld
Both amusing and scary as shit! If You’re Not Paranoid, You’re Crazy: The Atlantic on #surveillance www.theatlantic.com/magazine/archive/2015/11/if-youre-not… via @DavidBrin
-
Daniel Weld
@dsweld
Neat use of LSTM sequence-to-sequence mapping: auto email replies from ur phone
googleresearch.blogspot.com/2015/11/computer-respond-to-t… -
Daniel Weld reposted
Madrona
@MadronaVentures
We are celebrating our 20th year with support for STEM education via great orgs @codeorg @OppScholarship @geekwire www.madrona.com/madrona-code-wsos-geekwire/
-
Daniel Weld
@dsweld
Scary stuff in China: "authoritarianism, gamified" www.aclu.org/blog/free-future/chinas-nightmarish-citizen-…
-
Daniel Weld
@dsweld
Nice, hype-free article about the inherent risks of Artificial Intelligence www.microsoft.com/en-us/research/redirect/?ref=https://re… by @tdietterich and @erichorvitz
-
Daniel Weld reposted
Pioneer Square Labs
@psl
Hello, world! www.geekwire.com/2015/top-seattle-investors-raise-12-5m-f…
-
Daniel Weld reposted
Joe Hellerstein
@joe_hellerstein
Excited to see this @marcua and team bring this to reality!
B12
@b12
Orchestra is part of our goal to improve the future of work for creative & analytical people. Our recent blog post: blog.unlimitedlabs.com/post/130097884236/introducing-orch…
-
Daniel Weld
@dsweld
Congrats to UW CSE graduate Christopher Ré - just announced as a #macfellow Genius. www.macfound.org/fellows/943/
-
Daniel Weld
@dsweld
Congrats to @tomalberg on the TechImpact Lifetime Achievement Award! www.madrona.com?p=1303 Great recognition for @MadronaVentures
-
Daniel Weld
@dsweld
Insightful reflection on a decade of work at Google by @AlonHalevy & the contrast with academia: wp.sigmod.org?p=1851 via @adityagp
-
Daniel Weld reposted
Ed H. Chi
@edchi
Oh sh*t! workspaceupdates.googleblog.com/2023/04/new-community-fea…
-
Daniel Weld
@dsweld
Q&A with @pmddomingos about machine learning, "The Master Algorithm" www.washington.edu/news/2015/09/17/a-q-a-with-pedro-domin…
-
Daniel Weld
@dsweld
Lifelong education and nanodegrees. @SebastianThrun says college education won't carry you for a lifetime www.nytimes.com/2015/09/17/technology/udacity-says-it-can…
-
Daniel Weld
@dsweld
Computers may soon start programming themselves, says Cornell's Bart Selman read.bi/1P3gob5
-
Daniel Weld
@dsweld
Got my copy of @pmddomingos 's The Master Algorithm early - lucid explanation of Machine Learning! Great... www.amazon.com/Master-Algorithm-Ultimate-Learning-Machine…
-
Daniel Weld
@dsweld
Will an algorithm be your next boss? Great discussion on the future of work by @msbernst psmag.com/economics/the-future-of-work-working-for-the-ma…
-
Daniel Weld reposted
Panagiotis "Takis" Metaxas
@takis_metaxas
How Facebook’s privacy policy generally decreased in clarity and protections from 2005 to 2015
techscience.org/a/2015081102/ -
Daniel Weld
@dsweld
App-search, deep linking, and the post-app mobile experience - how Google is fighting back. on.recode.net/1fZ8lQ9
-
Daniel Weld
@dsweld
AshleyMadison: Turns out most users weren't adulterers after all (they just wanted to be)... x.com/etzioni/status/636752471520481280
-
Daniel Weld reposted
Medicalskeptic
@medskep
On numbers being gamed, people can't keep their own score - @EdwardTufte Read this carefully and often
-
Daniel Weld reposted
TechCrunch
@TechCrunch
Spare5 Raises $10M Series A Round For Its Mobile Mechanical Turk Service tcrn.ch/1Pwbfbh by @fredericl
-
Daniel Weld
@dsweld
Why Amazon doubled their percentage fee for Mechanical Turk workers john-joseph-horton.com/how-much-should-mturk-cost-if-your… via @etzioni
-
Daniel Weld reposted
Madrona
@MadronaVentures
Congrats @uwcse @lazowska on generous @Microsoft gift to get more CS grads in WA. Our students deserve this! www.geekwire.com/2015/microsoft-gives-10m-to-university-o…
-
Daniel Weld
@dsweld
Madrona raises $300M sixth fund, fueling Pacific NW entrepreneurs www.madrona.com/fundvi-300million/
-
Daniel Weld
@dsweld
The importance of a new building for UW CSE www.seattletimes.com/opinion/editorials/legislature-shoul…
-
Daniel Weld
@dsweld
UW CSE continues to grow AI/ML expertise as Sham Kakade joins the faculty! news.cs.washington.edu/2015/05/04/sham-kakade-joins-uw-co…
-
Daniel Weld
@dsweld
Why bother to interview someone, when your computer can do it automatically with voice analysis? www.wuwm.com/health-science/2015-03-23/now-algorithms-are… Via @algorithmia
-
Daniel Weld
@dsweld
Great piece about @MadronaVentures driving investment in Seattle startups www.geekwire.com/2015/the-mighty-madrona-why-one-vc-firm-… @geekwire
-
Daniel Weld reposted
TED Talks
@TEDTalks
"In all of human history, anytime someone has waged a war on science, science has won." @djrothkopf #ted2015
-
Daniel Weld
@dsweld
Great additions to the team at @MadronaVentures: www.geekwire.com/2015/f5-ceo-john-mcadam-isilon-co-founde…
-
Daniel Weld reposted
ROC HCI
@rochci
Congrats to @HenryKautz for being the inaugural Robin & Tim Wentworth Director of the Institute for Data Science. www.facebook.com/permalink.php?story_fbid=786163751468970…
-
Daniel Weld reposted
Cornell University
@Cornell
Greg Morrisett to return to Cornell as @CornellCIS dean news.cornell.edu/stories/2015/03/greg-morrisett-return-co…
-
Daniel Weld reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
Best political cartoon I've seen in a long time.
-
Daniel Weld reposted
Ed Lazowska
@lazowska
UW CSE Ph.D. alum Tessa Lau: One of “50 Women in Tech Dominating Silicon Valley” news.cs.washington.edu/2015/02/24/uw-cse-ph-d-alum-tessa-…
-
Daniel Weld
@dsweld
Interesting @bgurley post on Uber's move towards ride sharing: combinatorics & market dynamics. Great idea: linkd.in/16fPoVq
-
Daniel Weld reposted
Ed Lazowska
@lazowska
Remembering Gaetano Borriello news.cs.washington.edu/2015/02/01/remembering-gaetano-bor…
-
Daniel Weld
@dsweld
Great article (balanced for once) on how to avoid the risks associated with AI technology
medium.com/@tdietterich/benefits-and-risks-of-artificial-… -
Daniel Weld
@dsweld
"Why argue about the effect of AI on society; let's study it?" @erichorvitz gifts $ to Stanford: www.nytimes.com/2014/12/16/science/century-long-study-wil… via @jwnichls
-
Daniel Weld
@dsweld
Thought-provoking critique on stackoverflow's point system & community michael.richter.name/blogs/why-i-no-longer-contribute-to-…
-
Daniel Weld reposted
Johan Ugander
@jugander
Complementing discussions of the NIPS experiment, @karger has been tracking CHI review variability for past 3 years: haystack.csail.mit.edu/blog/2014/01/15/allocating-chi-rev…
-
Daniel Weld
@dsweld
Interview with @etzioni on the future of AI & the Allen Institute for AI www.nytimes.com/2014/12/16/science/paul-allen-adds-oomph-…
-
Daniel Weld reposted
Ed Lazowska
@lazowska
UW CSE dominates Allen Distinguished Investigator Awards in AI news.cs.washington.edu/2014/12/03/uw-cse-dominates-allen-…
-
Daniel Weld
@dsweld
Why Drugs Are a Metaphor for Some Foods, and Sex Is a Metaphor for Others www.newrepublic.com/article/119352/what-language-yelp-rev… via @tnr
-
Daniel Weld reposted
Ed Lazowska
@lazowska
Middle schoolers, high schoolers, and families: Attend UW’s Computing Open House, December 6 news.cs.washington.edu/2014/11/15/middle-schoolers-high-s…
-
Daniel Weld
@dsweld
You've seen big DATA... How about big DATING? Geek hacks okcupid, finds love... wrd.cm/1jsN6Ry
-
Daniel Weld
@dsweld
Cutthroat capitalism: a mole reveals Uber’s poaching tactics www.nytimes.com/2014/08/28/upshot/ubers-secret-agents-was…
-
Daniel Weld
@dsweld
Using ML, Bloomberg Ventures predicts who is a likely startup founder
techcrunch.com/2014/03/26/using-analytics-bloomberg-beta-… (via @djrosent) -
Daniel Weld
@dsweld
Check out Madrona's @JulieSandler pumping the crowd on girl power & entrepreneurship... Wow! vimeo.com/97890779?email_id=Y2xpcF90cmFuc2NvZGVkfGFkNzdhZ…
-
Daniel Weld reposted
Nate Silver
@NateSilver538
American political history since 1948, in one chart? 53eig.ht/1mM17Sb
-
Daniel Weld
@dsweld
Cool new iPhone app that selects the perfect still photo from a video feed www.geekwire.com/2014/meet-vhoto-new-iphone-app-scans-vid…
-
Daniel Weld reposted
Barney Pell
@barneyp
Great, rich article on Military, Law Enforcement and Intelligence Agency Chatbots www.eff.org/deeplinks/2014/04/answers-questions-about-mil… via @EFF
-
Daniel Weld
@dsweld
Vote for Oren Etzioni - AI2's hire of the year! www.geekwire.com/2014/geekwire-awards-vote-hire-year-2/
-
Daniel Weld
@dsweld
The anatomy of a $631k kickstarter campaign. Dan Shapiro shares tricks of the trade from Robot Turtles. www.geekwire.com/2013/anatomy-631230-kickstarter-campaign/
-
Daniel Weld reposted
Jure Leskovec
@jure
What You Get When You Pour Molten Aluminum Into An Ant Hill?
digg.com/video/what-you-get-when-you-pour-molten-aluminum… -
Daniel Weld
@dsweld
Thought provoking (and chilling). How shall we handle ubiquitous surveillance? @DavidBrin's take: www.davidbrin.com/transparentsociety1.html
-
Daniel Weld
@dsweld
Cool idea (cloud-based application lifecycle management) www.geekwire.com/2013/shippable-raises-2m/
-
Daniel Weld
@dsweld
Fascinating, useful and sweet... www.npr.org/blogs/thesalt/2013/12/03/248347009/cookie-bak…
-
Daniel Weld
@dsweld
Fossilized sex: www.nytimes.com/2013/12/01/sunday-review/the-shocking-sex…
-
Daniel Weld
@dsweld
Samsung's clever revenge on Apple: $1B fine paid in nickels via 30 dump trucks! bit.ly/1hTcHIu
-
Daniel Weld reposted
Ed Lazowska
@lazowska
Ben Taskar, 1976-2013 news.cs.washington.edu/2013/11/18/ben-taskar-1976-2013/ via @sharethis
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
Great video, nice viral referral incentive, urgency (50% off) - these guys "get" crowdfunding. www.hackthings.com/coin-one-card-to-rule-them-all/
-
Daniel Weld
@dsweld
Creative Live gets $21M for online education; interesting viral model: live content is free; you pay to access later
bit.ly/1idEkcX -
Daniel Weld reposted
Freakonomics
@Freakonomics
Pirate Economics, Somali Edition: numbers on pirate earnings & the financing and profit of pirate expeditions. freakonomics.com/2013/11/06/pirate-economics-somali-edition/
-
Daniel Weld
@dsweld
Depressingly fascinating: The Decline of Wikipedia via @jure technologyreview.com/featuredstory/520446/the-decline-of-…
-
Daniel Weld reposted
Mike Mathieu
@mikemathieu
.@walkscore's @mrlerner on How I taught my kids Python programming www.geekwire.com/2013/commentary-taught-kids-python-progr… #edtech
-
Daniel Weld reposted
𝐃𝐚𝐯𝐢𝐝 𝐁𝐫𝐢𝐧
@DavidBrin
Why We Live 40 Years Longer Today Than We Did in 1880: Chart from The Atlantic www.theatlantic.com/magazine/archive/2013/11/die-another-…
-
Daniel Weld reposted
Panos Ipeirotis
@ipeirotis
How Google taught itself good design. www.fastcodesign.com/3016268/google-the-redesign
-
Daniel Weld reposted
Barney Pell
@barneyp
NL and voice queries! Google unveils major overhaul of its search engine www.usatoday.com/story/tech/2013/09/26/google-overhauls-s…
-
Daniel Weld reposted
Hidden Brain
@HiddenBrain
MT @risatrix: @knowtheory Fascinating. From a UX/design perspective +1 on trash bin frowny faces. www.npr.org/2013/09/27/226580668/how-recycling-bias-affec…
-
Daniel Weld reposted
John Horton
@johnjhorton
Online Labor: Impressions from a visit to a large call center onlinelabor.blogspot.com/2013/09/impressions-from-visit-t…
-
Daniel Weld
@dsweld
Tinder transforms online dating with smart design & game mechanics ~ Instagram for dating: pandodaily.com/2013/08/26/laid-to-paid-how-tinder-set-fir…
-
Daniel Weld
@dsweld
OMG. You can no longer opt out of Facebook's Facial Recognition DB if you have a profile picture. singularityhub.com/2013/09/04/facebook-plans-to-add-milli… via @DavidBrin
-
Daniel Weld
@dsweld
Huge: Prof. Oren Etzioni leaves UW to head new AI institute funded by Paul Allen! www.geekwire.com/2013/microsoft-cofounder-paul-allen-taps…
-
Daniel Weld
@dsweld
I just backed Robot Turtles: The Board Game for Little Programmers on @Kickstarter kck.st/17BKz3h
-
Daniel Weld
@dsweld
New technology from an Israeli major general is upping the ante for facial recognition/creepiness: (cont) tl.gd/mji5m6
-
Daniel Weld reposted
Madrona
@MadronaVentures
Be Yourself, Redfin’s Glenn Kelman Says, Even if You’re a Little Goofy www.nytimes.com/2013/08/25/business/be-yourself-redfins-g…
-
Daniel Weld
@dsweld
Cool @PandoDaily Infographic: How will the MOOCs make money? pndo.ly/176pPSW
-
Daniel Weld reposted
-
Daniel Weld reposted
Laura Dabbish
@dabbish
Peer to peer food sharing? really? via @nprnews: A New App Will Let You Share Your Leftovers With Strangers www.npr.org/sections/alltechconsidered/2013/07/29/2064937…
-
Daniel Weld reposted
Larry Brilliant MD MPH FACPM 💛💙
@larrybrilliant
This is a stunning and important op-Ed re: charity and philanthropy by Peter Buffett, Warren Buffet's son. www.nytimes.com/2013/07/27/opinion/the-charitable-industr…
-
Daniel Weld
@dsweld
Surprising study of review fraud. Negative reviews come from fans! archive.nytimes.com/bits.blogs.nytimes.com/2013/07/15/why…
-
Daniel Weld
@dsweld
UW Prof Julia Parrish on another great example of citizen science: www.washington.edu/news/2013/07/10/julia-parrish-speaks-a…
-
Daniel Weld
@dsweld
America's Invisible laws. Scary stuff nytimes.com/2013/07/09/opinion/the-laws-you-cant-see.html via @mkapor
-
Daniel Weld
@dsweld
Amazing, but true: Lessons from the first millionaire online teacher pndo.ly/14YEEns RT @PandoDaily
-
Daniel Weld reposted
Pando
@PandoDaily
A (totally legitimate) letter from new Zynga CEO Don Mattrick pndo.ly/14Nklcd
-
Daniel Weld reposted
Freakonomics
@Freakonomics
More Evidence of the Link Between Obesity and Food Prices: A new working paper finds that the cost of food matters. bit.ly/1cqnIc9
-
Daniel Weld
@dsweld
Awesome - Poppy (@jheitzeb's 3D camera attachment for your iPhone) hits kickstarter goal in just 9 hours! kck.st/139jlns
-
Daniel Weld reposted
Pando
@PandoDaily
Poppy debuts on Kickstarter to turn you iPhone into a 3D camcorder pndo.ly/17Cjmlq
-
Daniel Weld
@dsweld
I just backed Poppy: Turn Your iPhone into a 3D Camera on @Kickstarter kck.st/12pALEG
-
Daniel Weld
@dsweld
Coffee & Power's 7 rules for virtual organizations - www.diamandis.com/the-launch-pad/creating-a-company-witho…
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
The smart watch category is getting crowded, but KREYOS looks pretty good - www.hackthings.com/kreyos-meteor-the-voice-and-gesture-co…
-
Daniel Weld
@dsweld
Analyzing statistical trends in twitter posts to detect health trends - great oped by @HenryKautz at www.nytimes.com/2013/06/23/opinion/sunday/theres-a-fly-in…
-
Daniel Weld
@dsweld
The rise & (surprising!) fall of automated high frequency trading. flip.it/141tT via @zang0
-
Daniel Weld
@dsweld
100 years ago today: incredible first ascent of 20,320' Denali. See this beautiful exposition: explore.americanalpineclub.org/index.php/simpleGallery/Sh…
-
Daniel Weld reposted
David Karger (hci.social/@karger)
@karger
Summarizing my keynote at the European Semantic Web Conference bit.ly/1b43tAl #eswc2013
-
Daniel Weld
@dsweld
These guys blow my mind - "Wi-Fi signals enable gesture recognition throughout entire home" www.washington.edu/news/2013/06/04/wi-fi-signals-enable-g…
-
Daniel Weld reposted
Brad Feld
@bfeld
When Patents Attack...Part Two! | This American Life www.thisamericanlife.org/radio-archives/episode/496/when-…
-
Daniel Weld
@dsweld
The beginning of the end for Coursera? higheredstrategy.com/coursera-jumps-the-shark/ via @rao_kambhampati
-
Daniel Weld reposted
Deep
@zang0
Imprisoned CIA Torture Whistleblower Describes Prison Life In Remarkable Detail flip.it/m8mzA
-
Daniel Weld
@dsweld
"Curing Robot Autism- A Challenge" - thought-provoking award paper at AAMAS-13 www.aamas-conference.org/Proceedings/aamas2013/docs/p801.pdf
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
3D Printing at Makerfaire was bigger than the entire On 3D Printing conference www.hackthings.com/3d-printing-at-maker-faire/
-
Daniel Weld
@dsweld
Buy burgers with cash. Your grocery purchases may raise insurance premiums... pandodaily.com/2013/05/20/you-are-your-data-the-scary-fut…
-
Daniel Weld
@dsweld
IRS audits 380,000 Romney supporters!!! archive.nytimes.com/fivethirtyeight.blogs.nytimes.com/201…
-
Daniel Weld
@dsweld
The new STDs - software transmitted diseases : licensing on Github. Via @Info_Activism readwrite.com/2013/05/15/open-source-is-old-school-says-t…
-
Daniel Weld
@dsweld
New apps and design guidelines for Google Glass archive.nytimes.com/bits.blogs.nytimes.com/2013/05/16/new…
-
Daniel Weld reposted
Josh Herst
@jherst
Google Maps redesign unveiled walk.sc/19uBQzF
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
iTunes for 3D Printing www.hackthings.com/itunes-for-3d-printing/ via @hackthingsblog
-
Daniel Weld reposted
Pando
@PandoDaily
Dept. of Homeland Security freezes accounts between Dwolla and bitcoin exchange Mt. Gox pndo.ly/10oidbc
-
Daniel Weld reposted
pourmecoffee
@pourmecoffee
“Huggies App Sends You a Tweet Whenever Your Kid Pees in His Diaper” www.adweek.com/adfreak/huggies-app-sends-you-tweet-whenev…
-
Daniel Weld reposted
pourmecoffee
@pourmecoffee
According to project by British TV, Shakespeare would have looked like a hipster today tollbit-host.com/culture/culturepicturegalleries/10030619…
-
Daniel Weld reposted
Meredith Ringel Morris
@merrierm
Fascinating article about constructing fake popularity (and thereby credibility) on Twitter: aka.ms/sn0dfo
-
Daniel Weld reposted
Josh Herst
@jherst
RT @geekestate: Walk Score’s New ChoiceMaps: A Deeper Measure of Walkable Neighborhood Amenities geekestateblog.com/walkscores-new-choicemaps-a-deeper-mea… Thx @SEATTLEHOMEoCOM
-
Daniel Weld reposted
John Horton
@johnjhorton
Great overview of Lant Pritchett's views on poverty & development: www.deseretnews.com/article/865576171/Everything-you-thin…
-
Daniel Weld reposted
-
Daniel Weld
@dsweld
Cynical, but thought provoking take on why MOOCs have momentum: bit.ly/XVX3D8
-
Daniel Weld
@dsweld
When freelance workers get ranked, the best do well. But do the others, victims of data darwinism, have any rights? bit.ly/ZTiX5y
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
Software eats hardware - www.hackthings.com/software-eats-hardware/
-
Daniel Weld
@dsweld
Great tools for someone who is starting a company on a shoestring! www.geekwire.com/2013/34-tools-launch-startup-idea-exit/
-
Daniel Weld
@dsweld
Guzdial asks interesting questions about MOOCs (esp about utility of forums) bit.ly/ZjiASf
-
Daniel Weld
@dsweld
Men think they can buy love?! http://
hrld.us/YETdYS via @HiddenBrain @spspnews: -
Daniel Weld
@dsweld
Interesting retrospective of TurKit (a foundational crowdsourcing IDE) crowdresearch.org/blog/?p=5006
-
Daniel Weld
@dsweld
IBM donates a version of Watson to a University (RPI ) onforb.es/11vqi0G
-
Daniel Weld
@dsweld
Provocative ideas about discourse and crowd creativity tom-atlee.posterous.com/the-wisdom-of-the-fringes via @draperdl
-
Daniel Weld reposted
Pando
@PandoDaily
A culture of lies and laziness pandodaily.com/2013/02/07/a-culture-of-lies-and-laziness/
-
Daniel Weld
@dsweld
Great illustration - the evolution of Google's design aesthetic under Larry Page: read.bi/YOzYBB
-
Daniel Weld
@dsweld
I love Seattle, but it's embarrassing that Washington's tax code is ranked as the most regressive in the US bit.ly/UKhgrj
-
Daniel Weld
@dsweld
Sad news about carbon: one x-country plane flight = a whole year of automobile driving www.nytimes.com
-
Daniel Weld
@dsweld
Google crowd sources a map of North Korea via @nytimestech www.nytimes.com/2013/01/30/world/asia/google-maps-new-tar…
-
Daniel Weld reposted
Henry Blodget
@hblodget
RT @Bill_Gross: "1/4 to 1/3 of all searches done on mobile are done with voice." Marissa Mayer of Yahoo at #wef
-
Daniel Weld reposted
Joe Heitzeberg
@jheitzeb
Great post today about building whimsical products starkravingvc.com/2013/01/25/calling-all-engineers-spice-… by @greggottesman
-
Daniel Weld reposted
Howard Moorey
@hojomo
Amazon Gets Into Voice Recognition, Buys Ivona Software To Compete Against Apple’s Siri: Amazon today announced ... feedproxy.google.com/~r/Techcrunch/~3/OgUvQebtZHo/
-
Daniel Weld
@dsweld
AirBnB, Lyft, Rover, SnapGoods & the rise (to $3.5B/yr) of the sharing-based economy onforb.es/XxuTa2
-
Daniel Weld
@dsweld
Stakes rise for "Review Wars" on Amazon - all legal per their terms of service :-( bit.ly/11MdeTx
-
Daniel Weld
@dsweld
Nice overview of Facebook search (my prediction: Google can relax) selnd.com/W9hlCQ
-
Daniel Weld
@dsweld
My new hero! Via @pourmecoffee “How 19-year-old activist Zack Kopplin is making life hell for Louisiana’s creationists” io9.com/5976112/how-19+year+old-activist-zack-kopplin-is-…
-
Daniel Weld reposted
Hidden Brain
@HiddenBrain
The many ways -- mostly unintentional -- that #gun #language seeps into our general conversation. www.nytimes.com/2013/01/16/us/gun-imagery-fills-language-…
-
Daniel Weld
@dsweld
California adopts Udacity to deliver courses for credit at San Jose State University-Interesting strategy for Udacity! www.nytimes.com
-
Daniel Weld
@dsweld
UW launches 8 startups in teh past few months bit.ly/XeD5wf
-
Daniel Weld
@dsweld
Interesting LA startup allows 'future location checkins" should G+ acquire? pandodaily.com/2013/01/09/google-is-missing-a-huge-opport…"
-
Daniel Weld reposted
𝐃𝐚𝐯𝐢𝐝 𝐁𝐫𝐢𝐧
@DavidBrin
No worries if you have nothing to hide? #smbc explores the boundaries of #surveillance and #privacy www.smbc-comics.com/index.php?db=comics&id=2849
-
Daniel Weld
@dsweld
Every hour of TV shortens your life by 22 min! New meta-study shows sitting is dangerous (despite regular exercise) bit.ly/13hX4jV
-
Daniel Weld
@dsweld
Captive Media raises $700k for urinal video games (advt biz model). Sigh. our last moments of quiet go "down the drain bit.ly/Wufl6l
-
Daniel Weld
@dsweld
Coursera gaining users even faster than Facebook! But can they make money? www.nytimes.com/2013/01/07/education/massive-open-online-…"
-
Daniel Weld reposted
Mark Reid
@mdreid
Out of curiosity, I grabbed gun death and gun ownership data for OECD countries from Wikipedia and ran it through R: dl.dropbox.com/u/38668/deaths-vs-guns.png
-
Daniel Weld
@dsweld
Good points about how Amazon is failing authors and publishers to the Dement if innovation. pandodaily.com/2012/12/28/i-admire-amazon-i-just-dont-sho…" via @panodaily
-
Daniel Weld
@dsweld
Want to know what it's like to work at an Amazon-style warehouse? m.motherjones.com/politics/2012/02/mac-mcclelland-free-on…
-
Daniel Weld reposted
Brad Stone
@BradStone
Santa tracks you all year then magically delivers stuff to your house. In other words, he's a mix of Facebook and Amazon's business models.
-
Daniel Weld
@dsweld
Fascinating analysis of emergent organization in crowdsourced journalism on Wikipedia gigaom.com/2012/12/19/what-wikipedia-can-tell-us-about-th… via @edchi
-
Daniel Weld
@dsweld
The _real_ reason Apple is bring manufacturing back to the US (provocative but not sure I buy it) www.asymco.com/2012/12/07/the-real-threat-that-samsung-po…
via @ipeirotis -
Daniel Weld
@dsweld
Great #nytimes article summarizing startup companies running specialized online labor markets www.nytimes.com
-
Daniel Weld
@dsweld
Good advice on choosing a VC from @greggottesman starkravingvc.com/2012/12/09/do-diligence-on-us/
-
Daniel Weld
@dsweld
Trailer for new Star Trek movie "Into Darkness" (May 2013) www.startrek.com/article/first-look-star-trek-into-darkne…
-
Daniel Weld
@dsweld
Prismatic raises $15M from Accel to perfect personalized news techcrunch.com/2012/12/05/prismatic/
-
Daniel Weld
@dsweld
Amazon changes the default to ensure high-quality workers on Mechanical Turk (but what makes a worker "master" level?) feeds.feedburner.com/~r/AComputerScientistInABusinessScho…
-
Daniel Weld
@dsweld
Gates Foundation invests $3M in MOOC course content and evaluation: bit.ly/RB8idg
-
Daniel Weld
@dsweld
Zombies are real! (just not in the human species) and the underlying biochemistry is really cool! www.nytimes.com
-
Daniel Weld reposted
Pando
@PandoDaily
Big companies take advantage of small Valley upstarts all the time, and VCs usually just walk away. Not this time. pandodaily.com/2012/12/05/there-is-a-lot-of-talk-about-be…
-
Daniel Weld
@dsweld
Can you trust reviews on Amazon? Their policy: "No need to experience product before review" archive.nytimes.com/bits.blogs.nytimes.com/2012/12/04/on-…"
-
Daniel Weld
@dsweld
A Lego tower 2 miles tall? www.bbc.co.uk/news/magazine-20578627 (via @jure)
-
Daniel Weld reposted
Steven Bird 🍉
@StevenBird
#nltk downloads now top 600 per day pypi.python.org/pypi/nltk
-
Daniel Weld
@dsweld
Staggering failure of social media to generate shopping activity (via @edchi) read.bi/Qk0fTU
-
Daniel Weld reposted
Ed H. Chi
@edchi
G+: Autistic workers might become the best and focused workers in your company soon. Software testers, for instance. … plus.google.com/110563789211679581171/posts/UmFev1Eohff
-
Daniel Weld reposted
Stephanie Clifford
@stephcliff
Hour by hour, how Amazon, Walmart & Target fight out online pricing. A1 tomorrow: www.nytimes.com/2012/12/01/business/online-retailers-rush…
-
Daniel Weld reposted
Panos Ipeirotis
@ipeirotis
Wonderful article by @jbreinlinger on how to bootstrap 2-sided markets (with examples): pandodaily.com/2012/11/20/liquidity-hacking-how-to-build-…





































































































































































































