AI:AM GUEST

Edward Hu

Head of AI Modeling, Mercor

Edward Hu is head of AI modeling at Mercor, where he has led model training and research since September 2026. At Microsoft Research he was first author of LoRA, one of the most widely used methods for fine-tuning large models, and a co-author of μTransfer. He paused his PhD with Yoshua Bengio at Mila in 2023 to help build OpenAI’s o1 reasoning model, and defended his thesis, “Building a Reasoning Machine,” in 2026.

APPEARANCES

One AI:AM appearance.

EPISODE 2026-10-07 · OCT 7, 2026

AI’s Unowned Risks and the Data Behind Frontier Agents

Evan Miyazono on relaunching Atlas Ignota to find AI risks no institution owns, agent identity, and formal verification; Edward Hu on Mercor’s realistic benchmarks, reward hacking, and the data frontier labs want next. Nathan and Prakash discuss an agent-made music track, OpenAI’s 722 math results, rising bond yields, and approval fatigue.

GUESTS · Evan Miyazono, Edward Hu