AI for Atoms: How Periodic Labs is Revolutionizing Materials Engineering with Co-Founder Liam Fedus

By No Priors Podcast

Categories: AI, VC

Summary

Physics PhDs are dominating AI leadership because the field offers 10-100x higher leverage than traditional physics research, with founders like Liam Fedus leaving OpenAI (where he co-created ChatGPT) to apply foundation models to materials science—the next frontier after consumer AI.

Key Takeaways

1. High-energy physicists pivoted to AI en masse after the Higgs boson discovery because apparatus bottlenecks made physics research progress slower than AI breakthroughs, demonstrating how external constraints drive talent migration.
2. Google Brain's 2016-2017 era was a 'Cambrian explosion' of AI research with small teams and minimal compute, proving that diversity and entropy in research—not massive scale—drove foundational innovations like transformers and mixture of experts.
3. ChatGPT's launch was intentionally chosen over specialized bots (meeting bot, coding bot) because founding teams like John Schulman advocated for general-purpose over narrow solutions—a strategic bet on breadth over optimization.
4. The transition from post-training at OpenAI to foundation models for materials science signals a new AI frontier: moving from digital/language tasks to physical world applications where AI can directly impact chemistry and atoms.
5. Physics backgrounds provide 'high leverage' in AI because physicists are 'very principled,' 'hard-nosed scientists,' and 'very careful'—character traits that translate to AI safety and rigorous system design rather than just technical skills.

Topics

• Foundation Models for Materials Science
• Post-Training at Scale
• Physics-to-AI Career Transitions
• AI Architecture Sparsity
• ChatGPT Product Development

Transcript Excerpt

Today on no prior we're talking with Liam Fetis. Liam is one of the co-creators of chat GPT which I think almost everybody uses at this point. He was the VP of post training at OpenAI and before that was at Google Brain where he worked on a variety of really early AI innovations. Liam will be telling us a bit about periodic labs his company which is focused on building an AI foundation lab for atoms. In other words, how do we impact the physical world, material sciences, chemistry, etc. using AI...