Two MIT Scientists Win 2026 Pritzker Prize for AI-Discovered Antibiotics

What did they discover?

The story starts in 2020, when Barzilay, Collins and then-MIT postdoc Jonathan Stokes — first author who worked in the Collins lab from 2017 to 2021 — published a landmark paper in the journal Cell. It described halicin, the first antibiotic identified by artificial intelligence.

The name is a nod to HAL 9000, the fictional AI in Stanley Kubrick's 1968 film 2001: A Space Odyssey.

Halicin's origins were unexpected. It had originally been studied as a potential diabetes drug. The researchers found it by training a deep learning model — an algorithm that learns patterns from data — on just 2,500 molecules. The model flagged halicin, and testing confirmed it killed antibiotic-resistant bacteria including Clostridium difficile, Acinetobacterium baumannii, and Mycobacterium tuberculosis.

A process that once took years took a few weeks.

In 2023, the team struck again. Working again with Stokes, now an assistant professor at McMaster University, they used machine learning to identify abaucin, a narrow-spectrum antibiotic effective against A. baumannii — a resistant bacterium commonly found on hospital doorknobs and in the wounds of U.S. soldiers stationed in Iraq and Afghanistan.

Narrow spectrum is the point. Because abaucin targets a single bacterial species, it lowers the odds that bacteria develop resistance and spares beneficial gut bacteria that broad-spectrum drugs would wipe out.

What do the laureates say?

Collins frames the work as a fundamental shift in how drugs are found. "Historically, antibiotic discovery has depended on laborious, trial-and-error strategies," he said. "This process is slow, costly, and unpredictable, often yielding few viable antibiotic candidates."

That's changing, he argues. "With generative AI, we're now seeing a shift from AI being used as a tool for discovery to AI being used as a design tool. This shift opens new frontiers in antibiotic discovery that could bring us the next generation of lifesaving drugs."

Barzilay is more cautious. "Despite significant progress in molecular modeling, we are still in early stages of drug discovery," she said. "With AI, we can only automate parts of the pipeline, rather than totally transforming the discovery process."

Her prescription for progress: "To make it happen, we need to work really hard to build realistic AI models of human biology."

Why it matters

Neither halicin nor abaucin is yet a marketed drug, and the laureates themselves say the field is young. But the pair have demonstrated a repeatable method: train a model on a modest set of molecules, screen for candidates, and validate in the lab. Against a resistance crisis that keeps outpacing traditional drug discovery, that proof of concept is the prize's real significance.