Nearly 2,000 AI agents spent 32 minutes investigating genetic changes across millions of mouse brain cells, producing possible leads for neurological disease research.

OpenAI, PerturbAI, and Scripps Research used the agents to analyze an existing dataset covering 7.7 million mouse brain-cell nuclei. The project generated interpretations linking gene disruptions to changes in cellular activity, including findings involving genes associated with rare neurological disorders.

The results suggest that AI could dramatically accelerate the analysis of complex biological datasets. But generating scientific hypotheses is not the same as validating them, and the proposed disease connections still require laboratory investigation.

Researchers gave each AI agent a separate gene investigation

Data came from a whole-brain CRISPR atlas containing measurements from 7.7 million mouse brain-cell nuclei.…