Artificial intelligence is transforming the search for treatments against some of medicine’s most stubborn foes, including Parkinson’s disease and antibiotic-resistant superbugs. Scientists say the technology is already producing promising new compounds that could lead to breakthroughs where traditional methods have stalled for decades.
One major area of progress involves the fight against drug-resistant bacteria. Each year, around 1.1 million people die from infections that were once easily treated, and that number could rise to more than eight million by 2050. Between 2017 and 2022, only 12 new antibiotics were approved, most of them similar to existing drugs that bacteria are already evading. Researchers at the Massachusetts Institute of Technology have now used a generative AI model to screen more than 45 million chemical structures. The AI identified two highly effective new compounds against Neisseria gonorrhoeae and Staphylococcus aureus, the bacteria behind gonorrhea and MRSA. Both compounds appear to attack bacteria in novel ways, raising hopes they could form a new class of medicines. The candidates are now undergoing further testing.
For Parkinson’s disease, which affects more than 10 million people worldwide, no treatment currently slows the disease’s progression. Scientists at the University of Cambridge trained a machine learning program on compounds known to target clumps of misfolded proteins called Lewy bodies, which are linked to neurodegeneration. The AI screened billions of potential molecules in days, a process that once took months and millions of pounds. After laboratory testing, the team identified five novel compounds that bind to Lewy bodies. “We can make very accurate predictions about the way candidate molecules will bind to the target at a scale that was unthinkable until a few years ago,” said Michele Vendruscolo, a professor at Cambridge and co-director of the Centre for Misfolding Diseases.
Researchers emphasize that AI does not replace traditional science but accelerates it dramatically. The technology can learn from its own mistakes by incorporating lab results into future searches, continuously improving its predictions. While the compounds discovered so far still require years of clinical testing, the speed and novelty of the findings offer a hopeful path forward for patients with conditions long considered untreatable.