GOLLuM trains LLMs using Bayesian uncertainty signals, raising high-yield Buchwald-Hartwig discovery from 24% to 43% in 50 experiments across 19 benchmarks.

Researchers have found a way to easily change one kind of atom into another in certain molecules, which are like building blocks for drugs.

Researchers in computational protein design and protein structure prediction assisted by AI have won the Nobel Prize in Chemistry.

Scientists discover highly conductive organic molecule to create smaller, more efficient, and powerful computing devices.

A predictive framework combines energy preferences with atom movement rates to forecast complete reaction sequences and guide better synthesis recipes for useful compounds.

This AI works as a chemical language model trained to predict chemical compounds that are useful for pharmaceutical applications.

Scientists at UT Dallas have developed a precise method to create "mirror molecules" for treatment of cancer and infections.

This discovery of a stable silicon version of an aromatic compound could open new paths to advanced materials and catalysts in chemistry and industry.

Artificial intelligence combined with a statistical technique reduces the heavy computing cost of high-precision electron-level calculations while preserving accuracy for bigger systems.

Scientists precisely attach organic rings that hold metals to narrow carbon strips, creating systems that combine magnetism, light emission, and sensing for quantum applications.