Liver fibrosis claims over 1.4 million lives annually, prompting Stanford geneticist Gary Peltz to enlist Google DeepMind's Co-Scientist AI to identify repurposable drugs. The tool analyzed existing literature and proposed three candidates, alongside Peltz's own selections, for laboratory testing against human liver cells.

Of the five candidates, Co-Scientist's suggestions outperformed Peltz's manual picks. Two AI-selected drugs blocked fibrosis and promoted liver cell regeneration, while none of Peltz's choices showed benefit. The standout was vorinostat, a cancer drug that blocked 91% of liver damage response. This gene-activity-targeting approach differs from traditional single-pathway treatments.

Peltz's findings suggest AI-driven drug repurposing could accelerate new anti-fibrotic therapies. By analyzing vast literature pools, Co-Scientist identified overlooked candidates that human researchers might miss, potentially transforming how we approach liver disease treatment.