Scientists used AI to sort through 7,500 candidate molecules to pinpoint a medicine that kills a dangerous superbug. Scientists tested thousands of drugs against A. baumannii to see which slowed down or killed the superbug. From this data, the AI ‘learned’ which chemical structures worked and which didn’t. The AI then predicted the efficacy of 7,500 new chemicals. It took just 90 minutes to narrow them down to 240, which scientists tested in a lab. Declaring abaucin the most effective.