The Digital Press

MIT has developed CRESt, an AI platform designed to act as a collaborative research partner by integrating diverse scientific data to plan and execute experiments autonomously.

Why it matters: CRESt accelerates scientific discovery by combining literature, images, experiments, and human feedback to guide complex research efficiently.

The big picture: Traditional AI tools handle limited data types, while CRESt uses multimodal inputs and active learning to explore broader experimental spaces.

Stunning stat: CRESt evaluated over 900 chemistries and conducted 3,500 electrochemical trials in just three months, achieving record fuel cell catalyst performance.

Commenters say: Some appreciate the system’s collaborative approach but express concern about the overuse of the term “copilot” in tech branding.