Harvard physicist Matthew Schwartz builds AI toolkit for precise scientific computation

Harvard physicist Matthew Schwartz said in a post on Anthropic’s science blog that scientific work with AI faces an "impedance mismatch," arguing that treating large language models as human-style collaborators is not the best way to draw out their scientific potential. To address that issue, Schwartz built a toolkit designed for precise calculation in quantitative science. Anthropic’s Claude then identified links between the toolkit and more than 10 fields, including ecology and population genetics, based on the recurrence of similar calculations across disciplines. Schwartz also worked with domain experts to guide the AI as it explored those questions. The update was cited by Techub, which attributed the information to Schwartz’s article and Anthropic.