PerturbAI and OpenAI Launch AI Reasoning Agents for Biomedical Discovery

According to Genetic Engineering and Biotechnology News, PerturbAI, OpenAI and their collaborators shared results from a year-long effort aimed at developing an AI reasoning framework[2] that helps scientists interpret large experimental datasets.OpenAI says the team divided the analysis among thousands of AI agents powered by GPT‑Rosalind[3]. The company says the agents studied a dataset showing how gene disruptions affect different types of cells across the mouse brain[3].In PerturbAI's announcement, the company said human experts preferred the reasoning-agents' interpretations 75% of the time[4] in blinded comparisons with reports from single AI agents. OpenAI says that at the largest workload, the single agent cited only 11 references while the distributed agents cited more than 700[3].OpenAI says the agents found a potential connection between cholesterol regulation and neurodegeneration[3]. Tech Times notes that this lead and the others "are hypotheses, not confirmed findings"[1] and that the preprint has not yet been peer-reviewed[1]. The outlet also reports that eight of the fourteen preprint authors are employees of either OpenAI or PerturbAI[1].PerturbAI CEO Grace Zheng said that "Biomedical experiments are generating data faster than scientists can interpret it"[4].According to Genetic Engineering and Biotechnology News, PerturbAI and OpenAI will now apply their AI reasoning frameworks to discover new drug targets[2] using PerturbAI's causal in vivo data. Joy Jiao, OpenAI's head of life sciences, said "this is just the beginning of what we can achieve together."[4]