The workshop will take place at Carnegie Mellon University, under the auspices of the Institute for Complex Social Dynamics and the Center for Formal Epistemology, and our new Center for Conceptual Research on Safe AI (CoRe Safe AI) on the weekend of September 19th to 20th, 2026.
This interdisciplinary meeting explores how formal methods from game and decision theory and the notions of agency and rationality from those contexts can help us to interpret, explain, and predict the behavior of frontier artificial intelligence. Sessions will cover open questions and novel results in AI agency and interpretability, belief elicitation, preference elicitation, and instrumental convergence.
Speakers
Speakers at the workshop include:
- Daniel Herrmann (Philosophy, University of North Carolina, Chapel Hill)
- Ben Levinstein (Alignment, Anthropic)
- Vincent Conitzer (Computer Science, Carnegie Mellon University)
- Aran Nayebi (Machine Learning, Carnegie Mellon University)
- Bryan Wilder (Machine Learning, Carnegie Mellon University)
- Aydin Mohseni (Philosophy, Carnegie Mellon University)
- Sven Neth (Philosophy, University of Pittsburgh)
Dates & Times
Saturday, September 19th, 9:00am–5:00pm
Sunday, September 20th, 9:00am–3:00pm
Location
Baker Hall A36, Adamson Wing, Carnegie Mellon University
Registration
Please join us! Attendance is free, but please register if you intend to attend.