The paper "A qualitative physics based on confluences" [18] by John Seely Brown and myself was one of the first papers in what is today a large and active area of artificial intelligence. As John Seely Brown has become the Chief Scientist of Xerox Corporation and is therefore somewhat distracted, this brief note represents my (Johan) personal views. 1 I will focus on some of the events and experiences which motivated me to become interested in qualitative physics and where it should be going.
When I was first exposed to the notions of symbolic computation and reasoning, these notions appeared as the ideal tools for the enterprise of understanding the physical world around us. Computers were already in widespread use in science and engineering for data analysis and simulation. However, even with the AI technologies of the 1970s, it was clear we could exploit computers in a far grander way. In particular, it became thinkable that we might be able to build systems that would be capable of reasoning about the physical world much as we ourselves, as engineers and scientists, do. I found this challenge enormously exciting because I saw (and still see) this as probably the only way that we will be able to build intelligences that can function successfully in the real world. Therefore, with the hubris characteristic of researchers in artificial intelligence, I set myself on a research programme with the goal of constructing what could best be described as an "artificial engineer" or "artificial scientist".