Robot chemist uses AI to discover new molecules

Supplementary content information

A robotic system which makes use of artificial intelligence (AI) to learn from chemical reactions and discover new molecules as a result, has been developed by chemists at the University of Glasgow, with support from the Engineering and Physical Sciences Research Council (EPSRC) and the European Research Council. The system uses AI and machine-learning to perform autonomous investigations on many reactions between substances.

The robot works initially with a human chemist to learn from their expertise about how to identify a discovery. The AI then takes over and the robot predicts which experiments to do next.

Until now, the process by which new molecules are discovered by human chemists, has been time-consuming and unpredictable, as well as constrained by human limitations and procedures. The hope is that the results of this research may help reduce the cost of finding new molecules by speeding up the process of discovery. These new molecules could be used for the development of new drugs and chemical products including materials and polymers, as well as use in high tech applications like imaging.

The team, led by Professor Lee Cronin, the University of Glasgow's Regius Chair of Chemistry, tested the system by combining 18 different chemicals, to create around 1000 reactions for further exploration by the robot.

In the event, the system required only around 100 of these reactions, before it became able to predict new possible chemicals and reactions to explore, with an accuracy of over 80 per cent. The robot can perform over 35 reaction experiments in a single day - around ten times the number that a human chemist can typically carry out.

The research has been published in the journal Nature.

Reference: PN 53-18