Artificial Intelligence and Philosophy of Science from the 1990s to 2020

The first three sections of the paper are by Donald Gillies. Section 1 argues that artificial intelligence (AI) is indeed relevant to quite a number of issues in philosophy of science. This paper will focus on one such issue – the problems of induction, confirmation and probability, which are indeed of key importance in philosophy of science. Section 2 traces the development of AI from the early 1970s to the 1990s, and section 3 examines the implications of this development for philosophy of science as set out in Donald Gillies’ 1996 book: Artificial Intelligence and Scientific Method. 24 years (nearly a quarter of a century) have elapsed since that book was published. During that time, there have been enormous advances in AI, and, specifically, in machine learning. The question therefore arises whether conclusions drawn in 1996 from the analysis of machine learning programs, which were state of the art at that time, still hold in the light of the much more powerful machine learning programs of 2020. In section 4, Marco Gillies, who uses contemporary machine learning in his research into virtual reality (VR), gives a brief account of some of the advances in machine learning since 1996. He goes on, in section 5, to examine how the conclusions stated in section 3 need to be modified in the light of advances in AI since 1996