Abstract

Machine discovery is a new area of artificial intelligence, dealing with computer systems which make discoveries. An automated discovery system can be imagined similarly to a human discoverer or to the sommunity of scientists-discoverers, as a robot that makes experiments and uses empirical data to develop theories. The author argues that construction of discovery system and theories of their functioning is a new and attractive program for the philosophy of science. He reviews the existing discovery systems and presents the theoretical schema emerging from their analysis. A large reference list provides directions for further studies