OpenAI's Proof That L = BPL

Tuesday, November 17, 2026 - 4:15pm to 5:15pm
Refreshments: 
at 4:00 PM
Location: 
32-G449 (Kiva/Patil)
Speaker: 
William Hoza (University of Chicago)
Biography: 
https://cs.uchicago.edu/people/william-hoza/

OpenAI recently announced a proof that L = BPL, a longstanding conjecture in complexity theory. Informally, what this means is that every randomized algorithm can be converted into a deterministic algorithm that does the same job, without increasing the amount of memory the algorithm uses by more than a constant factor. In this talk, I will present an overview of the proof.