Researchers at the Massachusetts Institute of Technology have developed an artificial intelligence (AI) model that was able to identify a new powerful antibiotic. They trained a deep neural network to predict molecules with antibacterial activity, without any assumptions of how drugs work and without chemical groups being labelled. The model was then used to screen a library of around 6000 molecules under investigation for human diseases. It was asked to predict which molecules would be effective against E.coli and to show the molecules that look different from conventional antibiotics. From the results offered by the model, the researchers selected about 100 molecules for physical testing. One of them proved to be a powerful antibiotic, named halicin. The researchers say this is the first time when AI has identified new antibiotics itself, without using any previous human assumptions.