Monthly Archives: December 2018

Success with Consciousness

Posted on by
16

What would success look like, when it comes to the question of consciousness?

Of course it depends which of the many intersecting tribes who dispute or share the territory you belong to. Robot builders and AI designers have known since Turing that their goal is a machine whose responses cannot be distinguished from those of a human being. There’s a lot wrong with the Turing Test, but I still think it’s true that if we had a humanoid robot that could walk and talk and interact like a human being in a wide range of circumstances, most people wouldn’t question whether it was conscious or not. We’d like a theory to go with our robot, but the main thing is whether it works. Even if we knew it worked in ways that were totally unlike biological brains, it wouldn’t matter – planes don’t fly the birds do, but so what, it’s still flying. Of course we’re a million miles from such a perfectly human robot, but we sort of know where we’re going.

It’s a little harder for neurologists; they can’t rely quite so heavily on a practical demonstration, and reverse engineering consciousness is tough. Still, there are some feats that could be pulled off that would pretty much suggest the neurologists have got it. If we could reliably read off from some scanner the contents of anyone’s mind, and better yet, insert thoughts and images at will, it would be hard to deny that the veil of mystery had been drawn back quite a distance. It would have to be a general purpose scanner, though; one that worked straight away for all thoughts in any person’s brain. People have already demonstrated that they can record a pattern from one subject’s brain when that subject is thinking a known thought, and then, in the same session with the same subject, recognise that same pattern as a sign of the same thought.  That is a much lesser achievement, and I’m not sure it gets you a cigar, let alone the Nobel prize.

What about the poor philosophers? They have no way to mount a practical demonstration, and in fact no such demonstration can save them from their difficulties. The perfectly human robot does not settle things for them; they tell it they can see it appears to be able to perform a range of ‘easy’ cognitive tasks, but whether it really knows anything at all about what it’s doing is another matter. They doubt whether it really has subjective experience, even though it assures them that it’s own introspective evidence says it does. The engineer sitting with them points out that some of the philosophers probably doubt whether he has subjective experience.

“Oh, we do,” they admit, “in fact many of us are pretty sure we don’t have it ourselves. But somehow that doesn’t seem to make it any easier to wrap things up.”

Nor are the philosophers silenced by the neurologists’ scanner, which reveals that an apparently comatose patient is in fact fully aware and thinking of Christmas. The neurologists wake up the subject, who readily confirms that their report is exactly correct. But how do they know, ask the philosophers; you could be recording an analogue of experience which gets tipped into memory only at the point of waking, or your scanner could be conditioning memory directly without any actual experience. The subject could be having zomboid dreams, which convey neural data, but no actual experience.

“No, they really couldn’t,” protest the neurologists, but in vain.

So where do philosophers look for satisfaction? Of course, the best thing of all is to know the correct answer. But you can only believe that you know. If knowledge requires you to know that you know, you’re plummeting into an infinite regress; if knowing requires appropriate justification then you’re into a worm-can opening session about justification of which there is no end. Anyway, even the most self-sufficient of us would like others to agree, if not recognise the brilliance of our solution.

Unfortunately you cannot make people agree with you about philosophy. Physicists can set off a bomb to end the argument about whether e really equals mc squared; the best philosophers can do is derive melancholy satisfaction from the belief that in fifty years someone will probably be quoting their arguments as common sense, though they will not remember who invented them, or that anyone did. Some people will happen to agree with you already of course, which is nice, but your arguments will convert no-one; not only can you not get people to accept your case; you probably can’t even get them to read your paper. I sympathised recently with a tweet from Keith Frankish lamenting how he has to endlessly revisit bits of argument against his theory of illusionism, one’s he’s dealt with many times before (oh, but illusions require consciousness; oh, if it’s an illusion, who’s being deceived…). That must indeed be frustrating, but to be honest it’s probably worse than that; how many people, having had the counter-arguments laid out yet again, accept them or remember them accurately? The task resembles that of Sisyphus, whose punishment in Hades was to roll a boulder up a hill it invariably rolled down again. Camus told us we must imagine Sisyphus happy, but that itself is a mental task which I find undoes itself every time I stop concentrating…

I suppose you could say that if you have to bring out your counter-arguments regularly, that itself is some indicator of having achieved some recognition. Let’s be honest, attention is what everyone wants; moral philosophers all want a mention on The Good Place, and I suppose philosophers of mind would all want to be namechecked on Westworld if Julian Jaynes hadn’t unaccountably got that one sewn up.

Since no-one is going to agree with you, except that sterling band who reached similar conclusions independently, perhaps the best thing is to get your name associated with a colourful thought experiment that lots of people want to refute. Perhaps that’s why the subject of consciousness is so full of them, from the Chinese Room to Mary the Colour Scientist, and so on. Your name gets repeated and cited that way, although there is a slight danger that it ends up being connected forever with a point of view you have since moved on from, as I believe is the case with Frank Jackson himself, who no longer endorses the knowledge argument exemplified by the Mary story.

Honestly, though, being the author of a widely contested idea is second best to being the author of a universally accepted one. There’s a Borges story about a deposed prince thrown into a cell where all he can see is a caged jaguar. Gradually he realises that the secrets of the cosmos are encoded in the jaguar’s spots, which he learns to read; eventually he knows the words of magic which would cast down his rival’s palace and restore him to power; but in learning these secrets he has attained enlightenment and no longer cares about earthly matters. I bet every philosopher who reads this story feels a mild regret; yes, of course enlightenment is great, but if only my insights allowed me to throw down a couple of palaces? That bomb thing really kicked serious ass for the physicists; if I could make something go bang, I can’t help feeling people would be a little more attentive to my corpus of work on synthetic neo-dualism…

Actually, the philosophers are not the most hopeless tribe; arguably the novelists are also engaged in a long investigation of consciousness; but those people love the mystery and don’t even pretend to want a solution. I think they really enjoy making things more complicated and even see a kind of liberation in the indefinite exploration; what can you say for people like that!

AI turns the corner?

Posted on by
15

Is the latest version of AlphaZero showing signs of human style intuition and creativity?

The current AlphaZero is a more generalised version of the program, produced by Demis Hassabis and his team, that beat top human players of Go for the first time. The new version, presented briefly in Science magazine, is able to learn a range of different games; besides Go it learned chess and shogi, and apparently reached world-class play in all of them.

The Science article, by David Silver et al, modestly says this achievement is an important step towards a fully-general game-playing program, but press reports went further, claiming that in chess particularly AlphaZero showed more human traits than any previous system. It reinvented human strategies and sacrificed pieces for advantage fairly readily, the way human players do; chess commentators said that its play seemed to have new qualities of insight and intuition.

This is somewhat unexpected, because so far as I can tell the latest version is in fact not a radical change from its predecessors; in essence it uses the same clever combination of appropriate algorithms with a deep neural network, simply applying them more generally. It does appear that the approach has proved more widely powerful than we might have expected, but it is no more human in nature than the earlier versions and does not embody any new features copied from real brains. It learns its games from scratch, with only the rules to go on, playing out games against itself repeatedly in order to find what works. This is not much like the way humans learn chess; I think you would probably die of boredom after a few hundred games, and even if you survived, without some instruction and guidance you would probably never learn to be a good player, let alone a superlative one. However, running through possible games in one’s mind may be quite like what a good human player does when trying to devise new strategies.

The key point for me is that although the new program is far more general in application, it still only operates in the well-defined and simple worlds provided by rule-governed games. To be anything like human, it needs to display the ability to deal with the heterogenous and undefinable world of real life. That is still far distant (Hassabis himself has displayed an awareness of the scale of the problem, warning against releasing self-driving cars on to real roads prematurely), though I don’t altogether rule out the possibility that we are now moving perceptibly in the right direction.

Someone who might deny that is Gary Marcus, who in a recent Nautilus piece set out his view that deep learning is simply not enough. It needs, he says, to be supplemented by other tools, and in particular it needs symbol manipulation.

To me this is confusing, because I naturally interpret ‘symbol manipulation’ as being pretty much a synonym for Turing style computation. That’s the bedrock of any conventional computer, so it seems odd to say we need to add it. I suppose Marcus is using the word ‘symbol’ in a different sense. The ones and zeroes shuffled around by a Turing machine are meaningless to the machine. We assign a meaning to the input and so devise the manipulations that the output can be given an interpretation which makes it useful or interesting to us, but the machine itself knows nothing of that. Marcus perhaps means that we need a new generation of machines that can handle symbols according to their meanings.

If that’s it, then few would disagree that that is one of the ultimate aims. Those who espouse deep learning techniques merely think that those methods may in due course lead to a machine that handles meanings in the true sense; at some stage the system will come up with the unknown general strategy that enables it to get meaningful and use symbols the way a human would. Marcus presumably thinks that is hopeless optimism, on the scale of those people who think any system that is sufficiently complex will naturally attain self-awareness.

Since we don’t have much of an idea how the miracle of meaning might happen it is indeed optimistic to think we’re on the road towards it. How can a machine bootstrap itself into true ‘symbol manipulation’ without some kind of help? But we know that the human brain must have done just that at some stage in evolution – and indeed each of our brains seem to have done it again during our development. It has got to be possible. Optimistic yes, hopeless – maybe not.

New paths to AI disaster

Posted on by
6

I’ve never believed that robots dream of killing all humans. I don’t think paperclip maximisers are ever going to rule the world. And I don’t believe in the Singularity. But is AI heading in some dangerous directions? Oh yes.

In Forbes, Bernard Marr recently offered five predictions for the year ahead. They mostly strike me as pretty believable, though I’m less optimistic than he is about digital assistants and the likelihood of other impressive breakthroughs; he’s surely right that there will be more hype.

It’s his first two that prompted some apprehension on my part. He says…

  1. AI increasingly becomes a matter of international politics
  2. A Move Towards “Transparent AI”

Those are surely right; we’ve already seen serious discussion papers emerging from the EU and elsewhere, and one of the main concerns to have emerged recently is the matter of ‘transparency’ – the auditability of software. How is the computer making its decisions?

This is a legitimate, indeed a necessary concern. Once upon a time we could write out the algorithm embodied in any application and check how it worked. This is getting more difficult with software that learns for itself, and we’ve already seen disastrous cases where the AI picked up and amplified the biases of the organisation it was working for. Noticing that most top executives were white middle-aged men, it might decide to downgrade the evaluation of everyone else, for example. Cases like that need to be guarded against and managed; it ought to be feasible in such circumstances, by studying results even if it isn’t possible to look inside the ‘black box’.

But it starts to get difficult, because as machine learning moves on into more complex decision making, it increasingly becomes impossible to understand how the algorithms are playing out, and the desired outcomes may not be so clear. In fact it seems to me that full transparency may be impossible in principle, due to human limitations. How could that be? I’m not sure I can say – I’m no expert, and explaining something you don’t, by definition, understand, is a bit of a challenge anyway. In part the problem might be to do with how many items we can hold in mind, for example. It’s generally accepted that we can only hang on to about seven items (plus or minus a couple) in short-term memory. (There’s scope for discussion about such matters as what amounts to an item, and so on, but let’s not worry about the detail.) This means there is a definite limit to how many possible paths we can mentally follow at once, or to put it another way, how large a set of propositional disjunctions we can hang on to (‘either a or b, and if a, either c, d, or e, while if b, f or g… and there we go). Human brains can deal with this by structuring decisions to break them into smaller chunks, using a pencil and paper, and so on. Perhaps, though, there are things that you can only understand by grasping twenty alternatives simultaneously. Very likely there are other cognitive issues we simply can’t recognise; we just see a system doing a great job in ways we can’t fathom.

Still, I said we could monitor success by just looking at results, didn’t I? We know that our recruitment exercise ought to yield appointments whose ethnic composition is the same as that of the population (or at any rate, of the qualified candidates).  OK, sometimes it may be harder to know what the desired outcome is, exactly, and there may be issues about whether ongoing systems need to be able to yield sub-optimal results temporarily, but those are tractable issues.

Alas, we also have to worry about brittleness and how things break. It turns out that systems using advanced machine learning may be prone to sudden disastrous failure. A change of a few unimportant pixels in a graphic may make an image recognition system which usually performs reliably draw fantastic conclusions instead. In one particular set of circumstances a stock market system may suddenly go ape. This happens because however machine learning systems are doing what they do, they are doing something radically different from what we do, and we might suspect that like simpler computer systems, they take no true account of relevance, only its inadequate proxy correlation. Nobody, I think, has any good theoretical analysis of relevance, and it is strongly linked with Humean problems philosophers have never cracked.

That’s bad, but it could be made worse if legislative bodies either fail to understand why these risks arise, or decide that on a precautionary basis we must outlaw anything that cannot be fully audited and understood by human beings. Laws along those lines seem very likely to me, but they might throw away huge potential benefits – perhaps major economic advantage – or even suppress the further research and development which might ultimately lead to solutions and to further, as yet unforeseeable, gains.

That’s not all, either; laws constrain compliant citizens, but not necessarily everyone. Suppose we can build machine learning systems that retain a distinct risk of catastrophic failure, but outclass ordinary human or non-learning systems most of the time. Will anyone try to build and use such systems? Might there be a temptation for piratical types to try them out in projects that are criminal, financial, political or even military? Don’t the legitimate authorities have to develop the same systems pre-emptively in self-defence? Otherwise we’re left in a position where it’s not clear whether we should hope that the ‘pirate’ systems fail or work, because either way it’s catastrophe.

What on earth is the answer, what regulatory regime or other measures would be appropriate? I don’t know and I strongly doubt that any of the regulatory bodies who are casting a thoughtful eye over this territory know either.