To plan the syntheses of small organic molecules, chemists use retrosynthesis, a problem-solving technique in which target molecules are recursively transformed into increasingly simpler precursors. Computer-aided retrosynthesis would be a valuable tool but at present it is slow and provides results of unsatisfactory quality. Here we use Monte Carlo tree search and symbolic artificial intelligence (AI) to discover retrosynthetic routes. We combined Monte Carlo tree search with an expansion policy network that guides the search, and a filter network to pre-select the most promising retrosynthetic steps. These deep neural networks were trained on essentially all reactions ever published in organic chemistry.

Chemical synthesis is a scientific procedure that uses simple chemical compounds to construct complex ones. It is the exact process used to create most of the substances we use in our daily life – from drugs to those components inside our electronic devices. However, the steps involved are often time intensive and extremely complicated that it may take years for chemists to master or achieve breakthroughs in certain procedures. Nonetheless, artificial intelligence has come to the rescue. To make it simple for everyone we can say, artificial intelligence is the new lab technician for chemical synthesis.