MIT Artificial Intelligence Uncovers Powerful New Antibiotic
Scientists at MIT use an AI model to identify a novel antibiotic that kills highly resistant bacteria without triggering immediate resistance. The discovery marks a major step forward in the fight against dangerous superbugs.
Scientists at MIT use an artificial intelligence system to discover a powerful new antibiotic named halicin. Unlike conventional antibiotics that are slight variations of existing drugs, halicin kills bacteria by disrupting their electrochemical gradient, which prevents the organisms from producing energy-storing molecules.
The researchers achieve this breakthrough by training a neural network on 2,500 molecules, allowing the AI to automatically map molecular structures and predict their behavior. When the AI screens a library of 6,000 compounds, it identifies halicin as a highly effective treatment against resistant strains like E. coli, which shows no resistance to the drug after 30 days of testing.
While MIT successfully eradicates a common military infection in mice using halicin, human trials remain ongoing. The team already expands its search by screening over 100 million molecules in another database, identifying 23 additional candidates and exploring ways to design new antibiotics from scratch to combat dangerous superbugs.