Jeremy is a research scientist at Reflection AI who gained recognition in the AI community by achieving state-of-the-art scores on the ARC-AGI benchmark — 53.6% on ARC-AGI-Pub in 2024 and 29.4% on ARC-AGI-2, topping the leaderboard. His work focuses on evolutionary test-time compute and continual learning approaches to generalization.
Jeremy Berman appeared as a guest on ThursdAI, the weekly AI news podcast hosted by Alex Volkov. Browse the full guest directory or subscribe on Substack to never miss an episode.
ThursdAI — The weekly AI podcast. Every Thursday, live.
Subscribe Free →