Beren Millidge is introduced as Zyphra's chief technology officer developing open models in the September 2026 Dwarkesh panel on AI research.
Loading distillations…
Millidge argues that models can attempt harder goals than people can demonstrate when success is verifiable. Whether that learning transfers far enough to remove human guidance remains uncertain.
Open conversation, Beren Millidge and largely agree that choosing objectives and the next experiment remains difficult for models. They disagree about how long that limitation lasts and what could overcome it.
Follow extended conversations with researchers, builders and historians into specific claims and disagreements.
Interview statements belong to their speakers; an appearance does not establish expertise on every topic.
Public episode pages, audio, transcripts and an RSS feed; access varies by post