What will the future of agentic AI look like? It's not just tasks, but thinking, that will need to be supported.
@bilva_chandra and colleagues argue that "Future agents are likely to operate in environments that are populated by other autonomous epistemic agents, each with different owners, objectives and data sources, whom they will learn from and interact with" including envisaging the following categories of epistemic agents.
Full paper: arxiv.org/pdf/2603.02960
Reading Dostoyevsky and stumbled across this, which feels like a warning about AI chatbots from the past.
"And they will submit to us with cheerfulness and joy. The most agonising secrets of their consciences... all, all, will they bring to us. And we shall resolve it all. As they will attend our decision with joy, for it will deliver them from the great anxiety and fearsome present torments of free and individual decision. And all will be happy."
Ummmm...
From Brothers Karamazov, The Grand Inquisitor.




















































































































































































































































































































































































































