Artificial Intelligence in Chemistry

Abstract

The definition of artificial intelligence and the associated tasks of this branch of science are discussed. The tasks include pattern recognition, adaptation and learning, problem solving by means of expert systems or neural networks, and understanding the natural language and communication with a machine in it. The principles of problem solving are analyzed. It is demonstrated how artificial intelligence-based computer programs in which chemical expertise is encoded assist in structure elucidation, in the investigation of relations between structure and biological activity or chromatographic retention, etc.; problems emerging in the synthesis planning with a retrosynthetic analysis, or in the planning of experiments and intelligent consultations are dealt with. Several models used for structure elucidation and synthesis planning are evaluated. An overview is presented of additional expert systems which, along with artificial intelligence-based robotics, are used in intelligent instrumentation. Also discussed is the role of neural networks, which begin to be successfully employed in structure elucidation, synthesis planning, in intelligent instrumentation and in the treatment of natural languages. They are expected to be an important tool in the implementation of intelligent systems for the classification of chemical databases and prediction of properties of molecules.