AI and Philosophy: How Can You Know the Dancer from the Dance?

haranguing Sloman to drop the way he learned to do philosophy at Oxford and to start studying artificial intelligence instead. Two years later, Sloman presented his first paper at the International Joint Conference on Artificial Intelligence (IJCAI 1971). This paper, " Interactions between Philosophy and Artificial Intelligence: The Role of Intuition and Non-logical Reasoning in Intelligence, " in some ways extended his DPhil thesis, exploring the philosopher Immanuel Kant's ideas about the limitations of logical deduction and the non-analytic nature of some mathematical knowledge. But the paper primarily presented a philosophical critique of an influential AI paper written two years earlier—namely, " Some Philosophical Problems from the Standpoint of AI, " by John McCarthy and Patrick C. Hayes. Looking back, Sloman explains, " I tried to show how reasoning processes could be generalized to include manipulations of analogical representations that use relations to represent relations, instead of using only explicit symbols to represent relations, as in logic and algebra. " Coming of age in AI Sloman's philosophic investigation of analogical representations and reasoning continues to this day, but his conversion to AI was complete after a year's study on a research grant at the University of Edinburgh in 1972. " I learned more in that year in Edinburgh than in any year since about the age of four, " Sloman says. He returned to the University of Sussex in 1973, where he helped found the AI department. Five years later, he published The Computer Revolution in Philosophy (Harvester, 1978) in which he extolled AI's power to extend our ability to think. He also predicted that, within a few years, philosophers who weren't familiar with the main developments in AI could be " fairly accused of professional incompetence. " A design stance Throughout a long, productive academic and research career, Sloman has continued his philosophical work to free AI from the restrictions of particular formalisms or mechanisms and to promote instead a broad, flexible approach that might match its domain. " After I learned about AI, " Sloman says, " it soon became clear to me that an intelligent system would need an architecture combining multiple interacting components performing different tasks. " He wrote the first draft of an architectural specification during the year he spent at Edinburgh and subsequently expanded it as part of his book, The Computer Revolution in Philosophy (In the first of the book's two parts, Sloman …