Artificial Intelligence - ruminations

I had to write something about AI, because AI has a key place in the rationale behind what I am trying to do.

But I really don't know much about AI, certainly not AI as it is generally understood these days.

When I first became interested in AI (the first thoughts that I recall were back in 1967) a major theme in AI research was the automation of reasoning in first order predicate logic. This took a heavy knock along with the rest of AI, in the UK, when the Lightfoot report said it was all a waste of time. In the UK AI was then subject to increased influence by soft scientists, since AI was still respectable as research in Psychology, if not in Computer Science. Meanwhile, anyone really interested in logic went into Theoretical Computer Science or ended up in Formal Methods. At least that's the way it seemed to me, but I was never an academic and my view has been from a distance.

It appears to have been not just a UK but a worldwide phenomenon, so blaming the Lightfoot report is probably wrong. I'd be interested to know what really happened.

Anyway, when I decided to start this page I thought I had better check out on the Web what is happening now. This is what I discovered.

First, Jonathan Bowen's page at Oxford is probably the best collection of links in AI, just as it is in formal methods.

Second, in this as in other areas, the Web is long on links and short on content.

What I found in a (not very thorough) perusal of the AI links was, not just that the central topic of the sixties no longer figures at all, but that it has even been excised from the history. A couple of "AI timelines" I looked at seemed keen to bring into AI a whole host of happenings which preceded not only the term itself but also the invention of the digital computer. But contained no mention of the kind of work I am interested in.

I have on my bookshelf a two volume compendium on the Automation of Reasoning containing key papers over a couple of decades.

The center piece in these volumes is Robinson's work on resolution. The collection is balanced around the papers which precede and those which follow Robinson's work.

Robinson, however, to judge by the Web, has been exised from the history of AI.

The automation of reasoning is an area which seems to me central to both of the academic disciplines which are most relevant to my interests and to much of my work over the last couple of decades (1957-70). The two disciplines are formal methods and AI, each of which gives (IMHO) insufficient priority to research in the automation of reasoning in (more or less) classical logics.

UP HOME © RBJ created 1995/10/7 modified 1996/12/27