MIT researchers were able to create an AI system that was able to discover a type of antibiotic, titled halicin, which is able to protect against many different forms of resistant bacteria. The AI was able to select for different traits by identifying different molecular structures. The team of MIT researchers were able to train the AI using 2,500 different molecules, and 1,700 different kinds of drugs, as well as 800 different natural products. In total, the AI examined over 6,000 different compounds to find one that was effective.
The system has not yet been used in human trials, so it will not be implemented in medical practice yet. Following the initial success, the team of researchers used the same system to examine over 100 million other molecules.
Image Credit: courtesy of the Collins Lab at MIT