Skip to content

Using thousands of AI agents to uncover new leads in biological data

OpenAI

The system used GPT-Rosalind agents to analyze experimental data from millions of mouse brain cells in 32 minutes, generating a literature-linked reasoning atlas.

Key points

  • Human experts preferred distributed-agent reports over single-agent reports 75% of the time in blinded evaluations
  • Distributed agents cited more than 700 references at peak workload, compared to 11 cited by a single agent with the same dataset
  • The reasoning atlas and tooling have been made publicly available on GitHub

Why it matters

Biomedical researchers can rapidly synthesize massive cellular perturbation datasets with published literature to generate and prioritize testable laboratory hypotheses.

Sources