Sir James and the Robots - The Lighthill Report and UK's AI winter

Histories
1

In the (German) article “Sir James and the Robots” at HNF blog, we learn that he wrote a report in 1972 to the UK’s Science Research Council on the subject of AI prospects, with a severe negative effect on public funding and therefore progress.

The BBC screened a debate on this important issue:

This is the full video of the Lighthill debate on Artificial Intelligence, organized in 1973 to discuss the advances and limits of artificial intelligence, featuring James Lighthill, Donald Michie, Richard Gregory and John McCarthy.

You can read the report here:
http://www.chilton-computing.org.uk/inf/literature/reports/lighthill_report/p001.htm

A key phrase, which today doesn’t seem to worry us much although perhaps it should:

there will always be serious difficulties in establishing that any particular program must necessarily have an acceptable output of plans and decisions

A very short character sketch by HNF:

Lighthill was a specialist in calculations of aerodynamics and noise development of jet engines. He probably learned how to use computers early on. He was highly educated, multilingual and an eccentric. He often refused to pay police fines; when his train couldn’t stop at a station, he asked to be slowed down and just jumped out. His hobby was long-distance swimming in the open sea. On July 17, 1998, James Lighthill died of a heart attack while circumnavigating the island of Sark in the English Channel.

2 Likes
2

This is a fascinating presentation. What fascinated me most (still only about 2/3rds of the way) is the surprisingly poor level of argument on the part of all the speakers.

To start with, I was unhappy with Lighthill’s use of definitions, which borders on a no true scotsman approach. He speaks of general robots being a mirage, but never really defines what he means of “general”. Instead he talks about toy systems and the combinational problem, but at no point does he connected this to the term “general”, at least I never heard any definition. This sort of thing borders on the no true scotsman argument.

Then he speaks of those successes in AI, like chess programs, being through the use of heuristics, which he implies is a sort of cheating. But this neatly refutes his argument that there is no point in AI as a science because, by his own example, studying human solutions to problems does indeed provide a real advance. If AI is the intersection of study of human I and AI, then it seems like a success to me.

The next presenter, from UEdinburgh, presented a toy system and Lighthill dismisses his entire argument based on that. However, he makes a very important point (one I would hear McCarthy use a decade later) that one definition of “general” can indeed be met by changing the software. The same robot can be used to assembly cars or boats, and by implication, anything else. What is this if not general? However, he scuppers his entire argument by failing to point this out again at the end, there’s no “so by my understanding of generality, this robot is a general system because it can be so easily reprogrammed” (not that that IS the definition of general, but it is A definition of it), instead he simply concludes the demonstration and Lighthill simply dismisses it as an example of the toy system he was referring to.

McCarthy opens (and that’s where I had to stop) by suggesting that AI as a science is indeed a science of its own right. I think he would have done much better by directly attacking Lighthill’s claims as to what AI was and was trying to achieve - and no one in the world was better placed to define these terms, having actually done so!

I think he could have simply stated that no one in AI was attempting general AI in the short term, and most would consider that to be unrealistic. Thus, it is Lighthill that is talk about a toy model, a toy model of what AI was and was trying to achieve. If one accepts the critic’s definition of what something is, certainly we would expect it not to be met.

1 Like
3

I think, it may be worth pointing out 3 passages from the Lighthill Report (my subtitles):

  1. On the general perspective:

To supplement the important mass of specialist and detailed information available to the Science Research Council, its Chairman decided to commission an independent report by someone outside the AI field but with substantial general experience of research work in multidisciplinary fields including fields with mathematical, engineering and biological aspects. I undertook to make such an independent report, on the understanding that it would simply describe how AI appears to a lay person after two months spent looking through the literature of the subject and discussing it orally and by letter with a variety of workers in the field and in closely related areas of research.

  1. The identified core of AI

Thus, the whole case for the existence of a continuous, coherent field of Artificial Intelligence research (AI) depends critically on whether between categories A [Advanced Automation] and C [Computer-based Collaborative Neuroscience (CNS) studies] there exists a significant category of research that may be described as a Bridge category, B, as well as on the strength of the case for any researches in that category. The existence of research work in this category is hardly in dispute: such work, as stated earlier, has been voluminous for many years, but there are much greater difficulties in any attempt at clear identification of good reasons for putting resources into those researches. The activities and stated aims of work in category B are described in the remainder of section 2.

Here, letter B stands not only for Bridge activity, but also for the basic component of that activity: Building Robots. The whole concept of Building Robots is, indeed, seen as an essential Bridge Activity justified primarily by what it can feed into the work of categories A and C, and by the links that it creates between them.

  1. British perspectives and relevance

Research on AI in some other countries may be funded by military agencies (ARPA in USA) or by other mission-orientated public bodies. With this type of funding it is common for scientists to close their ranks and avoid public disagreement among themselves, in the hope that the total funds available for science may thus be enhanced to an extent that may outweigh any harmful results of a distribution of those funds determined on the basis of insufficient scientific discussion. Such optimism would be unjustified in a poorer country such as Britain, while the alternative approach here advocated accords with the desire to keep our AI research civilian expressed to the author by various British workers in the field. This suggests that decisions within the UK should be taken only after carefully contrasting and comparing different informed views of the research field’s future available to SRC. Thus, due weight should be given to the principle Heterarchy not Hierarchy (an AI maxim of considerable soundness concerned with file structures).

So it’s really about a layman’s perspective on the field, as contrived from the insights of an outsider of the field after 2 months of (privileged) research. This pretty much rules out any academic infights over definitions. Mind that McCarthy is still fighting for the validity of a discipline that is pretty much lacking a consistent framework. Lighthill’s approach to this problem is quite an empirical one, by cornering the problem, i.e., identifying established fields (advanced automation and CNS studies), which renders AI research pretty much a science of the gaps, or in more polite terms, a bridging discipline. Which also defines his core questions, as there are, is there enough of a gap and is there a realistic perspective for filling or bridging this gap in the next 25 years (which is the given horizon of the report). Finally, this will be a British perspective, a recommendation for the specific situation this country is in. – I think, it’s especially the latter part of this argument, which doesn’t allow McCarthy a pronounced statement.

McCarthy’s position in this discussion is a quite difficult one. He’s still collecting resources and problems that may turn out crucial to what may turn out as the problem, but can’t provide answers (the most influential achievement at that time is timesharing, which isn’t really specific to general robots). He isn’t really in a position to contribute to a definition in layman’s terms, while he is still staking his claim in academic terms. And, last but not least, he’s one of the lucky few riding on ARPA and other highly funded budgets, which is – as explicitly stated – not the British option. So this is also really difficult in political terms. He’s condemned to be polite and to be not too specific. And, I’m afraid, while he’s one of the most prominent heads in the global discussion, he hasn’t much to contribute to this specific one.