Can AI do Theory?
Specifically, are modern AI methods capable of advancing the state of the art on frontier research problems, particularly in mathematics and theoretical computer science?
It is undeniable at this point that this is true in some form. And yet, several questions remain:
* What kinds of research problems is (current) AI technology best suited to tackle? What are the best techniques to do so?
* How can AI augment our current methodologies or entirely redefine how we approach rigorous mathematical research?
* What are the limitations of large language models when it comes to formal reasoning and theorem proving?
We are excited to be co-organizing a
workshop at STOC TheoryFest 2026 (
June 27 in Salt Lake City) that will explore these questions through invited talks, a panel discussion, and a poster session.
We have a spectacular lineup of invited speakers sharing their perspectives on the future of the field:
In-Person Speakers: Scott Aaronson, Carina Hong, Mark Sellke, David Woodruff
Virtual Speakers: Sebastien Bubeck, Prabhakar Raghavan
Call for Posters: We are actively seeking poster submissions! (Deadline
May 29 EOD)
Looking forward to seeing many of you in Utah!
Organizers: Pritish Kamath, Pravesh K Kothari, Mariana Raykova, Abhradeep Guha Thakurta, Nikhil Vyas