A new AI method speeds up the discovery of better molecules for medicine and materials.
AI combines genetic algorithms and language models to find promising molecules faster than ever.
Deep Dive
Researchers developed MolLIBRA, a new AI framework that accelerates the discovery of useful molecules. It uses a genetic algorithm guided by two smart filters: an ensemble of predictive models and a language model that scores molecules based on a text description of the desired property. This approach dramatically reduces the number of costly lab simulations needed, achieving top performance on 14 out of 22 benchmark tasks for molecular optimization.
Why It Matters
This could drastically cut the time and cost of developing new drugs and advanced materials.