On Spiritual Machines - A Much More Sceptical Perspective

Hi All

I have been following this discussion with a growing sense of bemusement, in that I would have thought that Ken - of all people! - would want to inject a somewhat more sceptical note into the discussion than he does. The whole discussion seems to assume that we can dismiss the idea that there is some inherent difficulty in the idea of intelligent conscious machines. Where it is alluded to at all, it is dealt with by noting that machines can now beat top go players as well as chess masters. This is the famous Turing Test ploy - specify what it is that you think computers will never be able to do, and we will eventually get enough computational power and clever programming together to show that it can be done. Thus, you can't win, because whatever you specify is just a matter of enough computational power and clever programming at some future time! If you object that this depends on human ingenuity to set the whole thing up, we hypothesise that computers will become increasingly self-programming, and so will be able to do it for themselves.

I personally don't find any of this at all convincing. I know that this is not a 'cool' position to adopt, but I can't help it - my many years of trawling through fundamental issues in the philosophy of mind have impressed me above all else with how very different psychological categories and qualities are to physical ones, and how very special are the systems which exemplify them to any degree of sophistication. For one thing, they are highly self-organised biological systems. Computers are currently not at all that, but highly sophisticated artifacts, made in our own image in certain highly particular ways, in that they are set up to mimic explicit computation tasks, and do them very much faster. I would argue that much, if not most, of what makes us human is not to do with computational power, and to suppose that consciousness is a matter of having enough of it is an egregious mistake. For one thing, computation is about explicit inference, not the immediate relational situation of being aware. As every meditator knows, thoughts, whether in inferential trains or of a more random variety, actually get in the way of consciousness expansion. Why then would mimicking thoughts (even if only the propositional content of them) be a likely way to make a system conscious? There is a huge disconnect between what the AI crowd think a system needs to do to be a conscious being and what spiritual teachings typically involve. Why does this not really come up in the dialogues more than it does?

There is a lot more to say, some of which is in my published paper "Why Computers Will Never Be People", available here: crpit.com/confpapers/CRPITV37Price.pdf  

In it I outline a plausibly 'Wilberian' view on the matter. Ken, If you are reading this, could you let me know where you differ?

Blessings

Keith Price