Top AI researchers are leaving big tech labs to join Periodic Labs, a start-up focused on accelerating scientific discovery through AI-driven experiments.
More than 20 leading artificial intelligence researchers have left OpenAI, Google DeepMind, Meta and other major labs to join a new Silicon Valley start-up, Periodic Labs, in recent weeks. Despite turning down lucrative offers worth tens of millions, the scientists say they are pursuing a different vision for AI.
Periodic Labs, founded by former OpenAI researcher Liam Fedus and ex-Google DeepMind scientist Ekin Dogus Cubuk, is focused on using AI to accelerate discoveries in physics, chemistry and other sciences. “The main objective of A.I. is not to automate white-collar work,” Mr. Fedus said. “The main objective is to accelerate science.”
Backed with $300 million in seed funding from a16z and others, the company is building a lab in Menlo Park, Calif., where robots will run large-scale scientific experiments guided by AI systems. Unlike traditional chatbot models, the researchers believe progress will come from teaching AI to learn from real-world trials.
“This will not make the discovery on the first try, but it will iterate,” Dr. Cubuk said. “After lots of iteration, we hope to get there faster.”