OpenAI has deployed nearly 2,000 parallel agents for scientific research: The system analyzed 8 million brain cells in just 32 minutes, uncovering new clues about neurological diseases.

Beating AI News Flash: OpenAI has partnered with U.S. AI biotech firm PerturbAI and the Scripps Research Institute, deploying 1,946 GPT-Rosalind-based agents to analyze experimental data from around 8 million mouse brain cells in parallel, completing the task in just 32 minutes. The research team estimates that this work would take scientists months to a year to finish manually, including analyzing data and reviewing literature. The data comes from a prior gene-editing experiment, where scientists modified different genes to observe changes in brain cells. The AI split nearly 2,000 sets of gene perturbation results among different agents, which analyzed cellular changes, reviewed relevant papers, and identified overlooked connections. One key finding involves the DCTN1 gene, whose mutations are linked to a rare disease with Parkinson’s-like symptoms. The AI observed that when this gene was disrupted, genes responsible for cholesterol synthesis in brain cells became more active. Drawing on existing research, it proposed an explanation: the issue may not be insufficient cholesterol, but impaired intracellular transport that prevents cholesterol from reaching its required locations. If this hypothesis is validated, restoring normal cholesterol transport could become a new approach to protecting nerve cells. The AI also found that Tsc1 and Tsc2, two genes that typically co-regulate cell growth, have distinct roles in the cerebellum. Disrupting Tsc2 affected dozens of genes involved in neural signal transmission, while disrupting Tsc1 caused almost no changes. This provides more specific clues for studying the genetic disorder tuberous sclerosis complex, though the underlying mechanisms require further experimental validation.