MIT professors Regina Barzilay and James Collins have been awarded the 2026 Pritzker Prize for AI in Science Research Excellence. Their collaboration combines deep learning, bioengineering and experimental testing to identify possible antibiotics for bacteria resistant to drugs. The University of Chicago and Caltech launched the prize in 2024 to honour work advancing AI alongside the natural sciences or engineering.
The researchers published their first breakthrough paper together in Cell in 2020, with Jonathan Stokes as first author. It described halicin, identified by a deep learning model trained on 2,500 molecules. Previously investigated as a potential diabetes treatment, halicin proved effective in tests against several antibiotic-resistant bacterial species. A discovery process that had taken years was completed within weeks.
Their collaboration with Stokes subsequently produced abaucin in 2023. Whereas halicin is a broad-spectrum antibiotic, abaucin targets a single species, Acinetobacter baumannii. This narrow-spectrum activity is considered an advantage because it minimises the likelihood of resistance developing and reduces the risk of damage to beneficial gut bacteria.
Collins contrasts these advances with conventional antibiotic discovery, which has relied on laborious, costly and unpredictable trial-and-error methods that often produce few viable candidates. He argues that generative AI is shifting from a discovery tool towards a design tool, opening possibilities for a new generation of antibiotics.
Barzilay nevertheless cautions that drug discovery remains at an early stage despite progress in molecular modelling. AI can currently automate individual parts of the pipeline rather than transform the entire process. Achieving that broader change, she says, requires more realistic AI models of human biology.
