🔬 Automating Science: World Models, Scientific Taste, Agent Loops — Andrew White

The podcast explores the intersection of AI and science, particularly the automation of scientific discovery. Andrew White, co-founder of Future House and Edison Scientific, discusses his journey from academia to industry, highlighting the potential of AI agents in automating the cognitive processes of scientific research, such as hypothesis generation and experiment design. White argues that while AI can accelerate scientific progress, bottlenecks remain in areas like lab work and data interpretation. He also touches on the importance of "scientific taste" and human preferences in guiding research directions. The conversation covers the development of AI tools like ChemCrow and Cosmos, emphasizing the role of world models in scientific automation.