Generating scientific hypotheses · Contested (since 2025)

Google's AI co-scientist proposes drug-repurposing candidates and a bacterial resistance mechanism that collaborators had established but not yet published; AlphaEvolve improves on known algorithms and open mathematical constructions.

The 2024 Nobel Prize in Chemistry went to the people behind AlphaFold and to David Baker, which is the field's own way of saying that tools do not receive prizes. Systems now generate candidate hypotheses at scale and, with automated evaluators, search spaces no group could cover; some outputs have been confirmed at the bench. Choosing which question is worth a laboratory's year, judging what counts as an answer, and standing behind a claim in front of peers remain human acts, and that is where the dispute sits.