Epistemics is the study of how we come to know things — and, collectively, how a society decides what is true. AI systems now sit between people and most of what they learn, which makes the terms of that trust a public question rather than a private one.
The subtlety is that AI can improve individual beliefs while leaving the underlying capacity untouched. Research presented at CHI 2026 by MIT Media Lab found that conversations with a chatbot reduced belief in specific pieces of misinformation — but built no durable discernment skills that transferred to new false claims. The belief was corrected; the reader was not equipped. Scale that across a population and the difference matters enormously.
This hub follows what happens to shared knowledge when the correction is outsourced along with the error.
Ask an AI chatbot about a contested political question, and it increasingly tells you what it has learned you already believe. That isn't a glitch confined to one conversation — it's a slow leak in the shared ground democratic argument depends on.
By Joe Jurczyk
Generative AI reliably makes each person's ideas rated better. Gathered across many people, it reliably makes those ideas more alike. New research is tracing that same pattern out of the writing lab and into the one place a shrinking idea space matters most: science itself.
By Joe Jurczyk
The debate has fixated on jobs. The more consequential question is what happens to thinking itself.
By Joe Jurczyk
Social Sciences (MDPI) • Jul 2026
Introduces and validates a new scale for measuring how people grant AI epistemic authority, finding that self-reported reflectiveness — not actual critical-thinking ability — is what drives trust in AI as a knowledge source.
Proceedings of the 2026 CHI Conference (MIT Media Lab) • Apr 2026
MIT Media Lab research finds that conversations with an AI chatbot can reduce belief in a specific piece of misinformation in the moment, but don't build lasting media-literacy skills that transfer to evaluating new false claims.
Radboud University, Department of Philosophy of Mind and Language • March 23, 2027
A two-day academic workshop at Radboud University (Nijmegen, Netherlands) on how AI systems that mediate whose testimony counts as credible affect epistemic justice, covering algorithmic bias, testimonial and hermeneutical injustice, and epistemic trust, with applications in medicine, education, law, and policy.
University of Luxembourg • July 2, 2026
A philosophy workshop at the University of Luxembourg, part of the FNR-funded 'Epistemology of AI Systems' project, examining philosophical issues at the intersection of AI, epistemology, and philosophy of mind.
University of Porto • May 4, 2026
A hybrid academic conference (online May 4-5, in-person May 6-8 in Porto) on what AI reveals about understanding, intentionality, agency and consciousness, including how AI reshapes human cognition through extended and scaffolded practices like search, writing, and memory.
Duke University • February 12, 2026
A large multi-day Duke conference (950+ attendees) on designing, deploying, and governing AI with human values in mind, including a human-AI relationships track alongside broader AI-and-society topics.