Regina Barzilay is a distinguished professor of AI and Health at MIT's School of Engineering, and James Collins holds the Termeer Professorship of Medical Engineering and Science at MIT. They were awarded the 2026 Pritzker Prize for AI in Science Research Excellence for their pioneering work in AI‑enabled antibiotic discovery. The prize, created in 2024 by the University of Chicago in partnership with Caltech, honors contributions that advance AI together with the natural sciences or engineering. Barzilay and Collins co‑lead the Abdul Latif Jameel Clinic for Machine Learning in Health and are affiliated with MIT's Schwarzman College of Computing and School of Engineering. In 2020 they published a breakthrough paper in Cell with then‑postdoc Jonathan Stokes, reporting the first antibiotic discovered by artificial intelligence—halicin. Named after HAL 9000 from 2001: A Space Odyssey, halicin was originally screened as a potential diabetes drug. A deep‑learning model trained on 2,500 molecules identified halicin within weeks, and experimental tests showed it kills resistant bacteria such as Clostridium difficile, Acinetobacter baumannii, and Mycobacterium tuberculosis, compressing a process that traditionally takes years. The result highlighted AI’s ability to rapidly pinpoint new, effective antibiotics amid the global antimicrobial‑resistance crisis. Building on that success, Barzilay, Collins, and Stokes (now an assistant professor at McMaster University) used machine learning in 2023 to discover abaucin, a narrow‑spectrum antibiotic effective against A. baumannii, a pathogen commonly found on hospital door handles and in wounds of U.S. soldiers deployed in Iraq and Afghanistan. Abaucin’s narrow spectrum reduces the likelihood of resistance development and spares beneficial gut microbes. Collins remarked, “Historically, antibiotic discovery relied on labor‑intensive trial‑and‑error, which is slow, costly, and unpredictable. With generative AI we are shifting from a discovery tool to a design tool, opening new frontiers for lifesaving drugs.” Barzilay added, “Despite advances in molecular modeling, drug discovery is still in its early stages. AI can automate parts of the pipeline, but transforming the process requires realistic AI models of human biology.”
Review