An AI that listens like a colleague: chemists built a tool that turns a scientist's hunch into an enzyme design

Enzymes are nature's tiny workhorses — proteins that speed up the chemical reactions behind medicines, foods and detergents — and redesigning one usually means months of expensive trial and error. Chemist Zhongyue (John) Yang and former doctoral student Qianzhen Shao have built MutexaGPT, an AI tool described in Nature Computational Science that takes a researcher's plain-English hunch — 'make this enzyme work in the cold' — asks the missing questions, builds physics-based models, and returns a concrete design plan. The point, Yang insists, is the opposite of replacement: 'We are showing that with AI you can translate human intuition, which is often imaginative and incredible, to create a workable plan.' Practical targets include enzymes for lactose-free milk, detergents that clean in cold water, and cleaner drug manufacturing.

SDG 9 · Industry, innovation and infrastructure — Mechanism: tech innovation