I recently attended a talk by Joanna Bryson, more “to the public” rather than in an academic setting, though it was to an educated public. The first half was an overview of mentality in robots, and touched (obviously briefly) on action selection and a little longer on consciousness issues. Dennet got a mention or two, for example his Multiple Draughts Hypothesis: “Constraints arise, get revised, contribute to the modulation of behaviour, and leave traces in memory.” Fine; I’ve got one of his books on consciousness and hope one day to read it (actually, I don’t expect to). She did say though that she considered consciousness not entirely verbal, contra Dennet, so that’s sound on her part.
What struck me as most extraordinary though, was the absence even from that half-hour’s brief outline, of any mention of Global Workspace Theory. Why do I think it should have been mentioned? Is it because Murray Shanahan has included it into his direction vector for further research? Shanahan is a prominent player, and wise, well-informed and able, so his involvement wouldn’t hurt. Is it because GWT has demonstrated through LIDA, one of Stan Franklin’s team’s implementations, an ability to give its user the impression they are actually communicating with a human, in the process of doing a useful job usually done by a human? Surely the ability to pass a kind of Turing test albeit within a limited subject area, might make it worth mentioning? Or is it because Patricia Churchland, when introducing the modern approach to consciousness on Charlie Rose and Eric Kandel’s rather good series (each program so like my tutorials – but with slightly more famous people 😉 ), chose to base her account on GWT and Baars? Well look, if PC bases her five minute account on that, it’s worth a line and a mention in anyone’s half-hour resumé on that kind of subject! Even if for no other reason! Anyway, the word conscious is now used rather widely, apparently justified by no more than a GWT process.
Yes, I do feel it’s perfectly possible to make major contributions to AI without writing significant software. The style of my palaeontological contribution was for a long time of an armchair nature, and though I did end up doing more significant experimental computer investigations than 99% of dinobird palaeontologists, I still pride myself in exemplifying just how much the armchair approach can contribute to science. I’m happy to claim this can also apply in AI. However, having said that, AI talk is best complemented by vivid computerised demonstrations of special abilities. If Bryson thinks there are approaches better than GWT, why not emphasise it with convincing implementations? Dennet’s view above has for example been embodied in at least one GWT implementation and probably all of them, for a long time. I remember writing one myself. Alternative paradigms may still be good even if they haven’t been implemented, but the GWT’s implementations have only scratched the surface. I might say some other impressive AI implementations are rather specialist, and not really flexible or general enough to qualify as real AI. And that doesn’t include the “promising” electromechanical doll projects with moving facial features and suchlike that Bryson depicted in her talk, which I do not count as impressive.
I was not impressed with those or the omission of GWT. And I’d been unimpressed with Bryson before. Some years ago I’d recognised her at a conference, and, realising she was becoming a notable fixture around the place, thought I’d better establish contact with her. I was treated to what I’ve come to recognise as the characteristic instantaneous response from female academics from the USA when meeting me face to face: it was as though she couldn’t understand how I had the incredible cheek even to think someone like me was allowed to talk to her. But I thought that was what conferences were for! Indeed some are even called colloquia. Although I don’t go to many conferences (maybe one day I’ll explain why I now think they’re overrated), and when I do, Americans and women are both minorities… and also, perhaps like most people, to avoid misunderstandings and complications I probably have a slight tendency to chat casually more with my own gender, I can nonetheless remember four occasions at conferences where I tried to talk for the first time to some female US academic face to face. With Bryson and two others, I got exactly the same very rude brush-off, with, amazingly, no sense of embarrassment in doing it, as though it were absolutely normal. Those three were white. The third was black, and we had a perfectly reasonable academic discussion. (I remember trying to explain to her a mechanism in terms of a sheep dip, totally lost on someone brought up in NYC!)
The odd US male academic can be a bit proud face to face but it isn’t a problem and they will talk to you (though I’ve learned not to open the conversation with a cheery “What are you famous for?”, for which I once had such high hopes as an opening gambit). Of course, the really significant major living academics involved with all areas of my work are usually from the USA, as are some of the really nice ones. I’ve got a feeling that actually US women are at least as likely if not more to reply to an email as the men. I should say it has become apparent to me over the years, that being (or becoming) a female academic in the USA can be tougher than in the UK/Europe in general. I get the impression there’s an issue not just with harassment but also with some institutions’ permissiveness or connivance with it. There’s something else too; I think there must be something to do with being taken seriously, which might be an extra burden, judging by the effect whatever it is has on them. Bryson’s own account of life in MIT did confirm my worst fears about college life in the USA; the same thing that “The Soul of a New Machine” had done for my view of working in commercial engineering over there.
In case she’d been preoccupied with something at the time, I subsequently emailed her inviting her to join a discussion list that a few people more significant than her were happy to join and participate in. I got not even a reply of course. Had she accepted she might have learned something potentially to the advantage of those who hope to tap her expertise. Instead she became an excellent example of rude ways to fail to do one’s duty as a Mad Postman [see first posting in this series] in scientific communication: the sacred duty to discover and pass on everyone’s ideas, whatever gloss you may put on them. For those scientists who consider themselves too important to be, amongst other things, an honest conduit, dredging is always a possibility.