AI Radar

Noteworthy AI conversations and writing, summarized from publisher notes.

AI for Atoms: Liam Fedus on Periodic Labs and Revolutionizing Materials Engineering

· No Priors

Liam Fedus, co-founder of Periodic Labs and former VP of post-training at OpenAI, discusses the application of AI to materials science and physical world problems. Drawing from his background in physics and AI research at Google Brain and OpenAI, Fedus explains how Periodic Labs aims to build an AI foundation lab for atoms, accelerating scientific discovery through closed-loop systems that integrate simulation, experimentation, and specialized neural networks. He highlights the challenges of data scarcity in physical sciences compared to language models and the importance of combining experimental data with simulations to improve model accuracy. Fedus also reflects on the interdisciplinary nature of the work at Periodic Labs, involving physicists, chemists, AI researchers, and engineers collaborating closely. He emphasizes the role of automation and robotics as accelerators for scaling experiments but notes that current hybrid human-robot systems are already producing valuable data. Looking ahead, Fedus envisions a future where AI-driven materials engineering dramatically speeds up physical innovation, akin to a new industrial revolution, and stresses the importance of connecting AI systems to the physical world to unlock transformative advances.

AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.