Keep pulling the thread on Liam and Doge.
Periodic Labs is a frontier AI research lab that uses Large Language Models (LLMs) to advance physics and chemistry.
Periodic Labs' core strategy involves tightly coupling physical experiments with simulations and LLMs to accelerate scientific R&D.
Periodic Labs uses its physical laboratory to create a "physically grounded reward function," where real-world experimental outcomes serve as the ground truth for training its AI models.
Doge believes that existing large language models are incapable of making novel scientific discoveries because they cannot iteratively conduct experiments.
Liam argues that an AI model's capabilities are specific to its training distribution, meaning a model trained on code will not generalize to solve problems like cancer research without the relevant data and iterative environment.
Periodic Labs' strategy is predicated on the belief that the necessary experimental data for their target scientific problems does not exist in public literature and must be generated in-house.
Periodic Labs' near-term commercial strategy is to build AI "co-pilots" for engineers and researchers in advanced industries such as space, defense, and semiconductors.
Periodic Labs' advisory board includes ZX Chen and Steve Kvelson from Stanford University, Mercury Kanazidis from Northwestern University, and Chris Wolverton.
Liam, a co-founder of Periodic Labs, was a co-creator of ChatGPT.
Doge, a co-founder of Periodic Labs, previously ran physics teams at DeepMind.
Early versions of ChatGPT were not mathematically strong because their initial reward functions were not optimized for mathematical correctness.
Periodic Labs is building a powder synthesis lab to create new materials, a method which can be automated by relatively simple robots.