AI Discovers New Drugs for Parkinson's and Superbugs

After 50 years of losing ground to drug-resistant bacteria, scientists just found a powerful new ally in an unexpected place: artificial intelligence.

Researchers at MIT used AI to discover two promising new compounds that can kill gonorrhea and MRSA bacteria that resist every other treatment. The technology screened 45 million chemical structures in days, a task that would have taken human scientists lifetimes to complete.

"We can look at massive libraries of chemical compounds in a matter of days or hours," says James Collins, professor of medical engineering at MIT. His team trained AI to recognize what makes antibiotics work, then let it design entirely new molecules from scratch.

The urgency couldn't be greater. Around 1.1 million people now die each year from infections that were recently easy to treat. Without new solutions, that number could climb to eight million deaths by 2050.

Between 2017 and 2022, only 12 new antibiotics received approval, and most resembled existing drugs that bacteria already resist. The AI-designed compounds work differently, attacking bacteria through entirely new pathways that could overcome their defenses.

But the technology isn't stopping at superbugs. Scientists are now pointing AI at Parkinson's disease, which has defeated treatment efforts for over 200 years since its discovery in 1817.

More than 10 million people worldwide live with Parkinson's, with rates climbing as populations age. In the UK, one in 37 people will receive a diagnosis in their lifetime. Despite countless clinical trials, no treatment can slow the disease's progression.

Michele Vendruscolo, professor of biophysics at Cambridge University, explains the challenge: "People are really confused about what the target should be. Even if you know the target, it's typically very difficult to go after it."

His team used AI in 2024 to identify potential compounds that could finally halt Parkinson's before it begins. The technology can analyze patterns in disease biology that human researchers might miss.

Collins' lab has also deployed AI against tuberculosis and C. difficile infections, discovering powerful new antibiotic candidates for both. Each success opens doors for thousands of rare diseases that lack any treatment options.

The Ripple Effect

The shift from traditional drug discovery to AI-powered research represents more than faster timelines. It fundamentally changes which diseases get attention and funding.

Rare diseases affecting small patient populations rarely attracted pharmaceutical investment because developing traditional drugs costs too much with too little return. AI dramatically lowers those costs, making previously "unprofitable" cures economically viable.

Drug-resistant infections disproportionately harm vulnerable populations in hospitals and nursing homes. New antibiotics designed by AI could save millions of lives in communities that have watched their treatment options vanish.

The technology also democratizes discovery. Smaller research teams can now accomplish what once required massive pharmaceutical budgets and decades of laboratory work.

What seemed impossible just five years ago is becoming routine: diseases once considered incurable are yielding to treatments invented by machines in days.