Descartes vs AI
Posted on 6th May 2009 at 10:18 by Clive Webster with 22 comments
Mark's excellent Future of AI article went up on the site recently, and reading it over it reminds me of my days back at Uni studying Philosophy, particularly René Descartes (1596 - 1650).
I'm not sure why I got into Descartes - most of modern philosophy is pedantic and dry to the point of making the reader narcoleptic - but Descartes was an interesting fellow. He was notoriously lazy, never getting up before midday, and spent most of the waking day in a dressing gown beside the fire reading. When he took a job to teach Queen Christina of Sweden he was forced to get up early to start his lessons however, and famously died of pneumonia due to the early starts.
The really interesting thing about Descartes (and the reason why I'm writing this) is because he would probably have believed that AI is impossible and that therefore anyone researching it was wasting their time.
The reason for this is Descartes' two fundamental beliefs about science and the human mind. He didn't just say, 'I think therefore I am' y'know...
Descartes stated that before one should invest time in scientific research, one should ascertain whether the subject being researched is possible. 'Possible' in the realms of philosophy is a tricky subject, but let's settle on defining it as 'logically possible' for now (otherwise we'll get into a discussion of Kripke's excellent Naming and Necessity, and that's too long a discussion for a blog).
Essentially, by saying that you should investigate the possibility of something being true before trying to prove whether it's true or not, Descartes is saying you must do your philosophy before you do your science.
He'd then move onto to his theory of the human mind to show that artificial intelligence is impossible. Descartes staunchly held the view that the human mind was a separate entity to the human body, a theory which is called Dualism.
The argument for any kind of Dualism is a complex one - to argue that a non-physical mind can control a physical body, and be affected by that physical body is far from a naive opinion.
Descartes' 17th Century Christian beliefs were an influence on his theory of Dualism, but it was mainly his pursuit of items of undeniable knowledge that resulted in Cartesian Dualism. To find such nuggets of incontrovertible truth he tried to doubt the truth of absolutely everything (Cartesian Doubt).
However, he found that he couldn't doubt the existence of himself as a thinking entity: as soon as you have a thought, you can't doubt that the existence of the thing having that thought - 'I think therefore I am.'
It's much easier to doubt the physical world, however. All our mental experiences are merely electrical signals delivered to the brain via nerves; if someone were to replicate these signals with a living brain sitting in a vat on some mad scientist's shelf, there would be no experiential difference.
That you could doubt the existence of your body but not your mind, Descartes said, meant that the mind and the body were (or at least, could be) distinct. It also gave the mind a certain superiority over the body and the base instincts of its 'animal spirits'.
Descartes never really provided a decent argument as to why such a superior entity as a mind would ever want to be attached to the human body, but his argument did mean that he believed anything purely physical couldn't think. Therefore, surely there can be no AI and so all these people playing around with robots are wasting their time. Descartes also believed that animals not only had no mind but were incapable of feeling pain, leading to live dissections of many animals. I'm guessing he wasn't a vegetarian.
Current-day philosophers have a very different view of the human mind however, which supports the work of the AI pioneers. The theory is Materialism and merely states that the human mind is merely a nice way of referring to a load of chemical and electrical happenings in the human brain, and no more. According to this theory, AI is eminently possible.
Materialism poses some serious questions for most theologians (or should that be the other way round?) and doesn't seem to capture the variation and wonder of the human mind. It seems rather a sad theory of mind compared to Descartes' fantastical ideas, doesn't it?
I'm not sure why I got into Descartes - most of modern philosophy is pedantic and dry to the point of making the reader narcoleptic - but Descartes was an interesting fellow. He was notoriously lazy, never getting up before midday, and spent most of the waking day in a dressing gown beside the fire reading. When he took a job to teach Queen Christina of Sweden he was forced to get up early to start his lessons however, and famously died of pneumonia due to the early starts.
The really interesting thing about Descartes (and the reason why I'm writing this) is because he would probably have believed that AI is impossible and that therefore anyone researching it was wasting their time.
The reason for this is Descartes' two fundamental beliefs about science and the human mind. He didn't just say, 'I think therefore I am' y'know...
The man himself
Descartes stated that before one should invest time in scientific research, one should ascertain whether the subject being researched is possible. 'Possible' in the realms of philosophy is a tricky subject, but let's settle on defining it as 'logically possible' for now (otherwise we'll get into a discussion of Kripke's excellent Naming and Necessity, and that's too long a discussion for a blog).
Essentially, by saying that you should investigate the possibility of something being true before trying to prove whether it's true or not, Descartes is saying you must do your philosophy before you do your science.
He'd then move onto to his theory of the human mind to show that artificial intelligence is impossible. Descartes staunchly held the view that the human mind was a separate entity to the human body, a theory which is called Dualism.
The argument for any kind of Dualism is a complex one - to argue that a non-physical mind can control a physical body, and be affected by that physical body is far from a naive opinion.
Descartes' 17th Century Christian beliefs were an influence on his theory of Dualism, but it was mainly his pursuit of items of undeniable knowledge that resulted in Cartesian Dualism. To find such nuggets of incontrovertible truth he tried to doubt the truth of absolutely everything (Cartesian Doubt).
However, he found that he couldn't doubt the existence of himself as a thinking entity: as soon as you have a thought, you can't doubt that the existence of the thing having that thought - 'I think therefore I am.'
It's much easier to doubt the physical world, however. All our mental experiences are merely electrical signals delivered to the brain via nerves; if someone were to replicate these signals with a living brain sitting in a vat on some mad scientist's shelf, there would be no experiential difference.
That you could doubt the existence of your body but not your mind, Descartes said, meant that the mind and the body were (or at least, could be) distinct. It also gave the mind a certain superiority over the body and the base instincts of its 'animal spirits'.
Descartes never really provided a decent argument as to why such a superior entity as a mind would ever want to be attached to the human body, but his argument did mean that he believed anything purely physical couldn't think. Therefore, surely there can be no AI and so all these people playing around with robots are wasting their time. Descartes also believed that animals not only had no mind but were incapable of feeling pain, leading to live dissections of many animals. I'm guessing he wasn't a vegetarian.
Current-day philosophers have a very different view of the human mind however, which supports the work of the AI pioneers. The theory is Materialism and merely states that the human mind is merely a nice way of referring to a load of chemical and electrical happenings in the human brain, and no more. According to this theory, AI is eminently possible.
Materialism poses some serious questions for most theologians (or should that be the other way round?) and doesn't seem to capture the variation and wonder of the human mind. It seems rather a sad theory of mind compared to Descartes' fantastical ideas, doesn't it?
22 Comments
Discuss in the forums ReplyA collection of chemical impulses arranged in such a fantastically improbable way as to produce complex thought, capable of reasoning and emotion, is a wonderfully interesting and exciting thing.
+1. Something different is always appreciated tbh. I'm liking these thoughts we're seeing dribbling down from the staff.
but I can't think of a good logical argument, I guess that's why I'm a geek not a philosopher.
Mathematically though the human brain is a Turing Machine, and as such can be emulated by any other Turing machine, so you can at least have an AI that is a copy of a human brain running on something other than grey matter, and if it talks like a person, and reasons like a person would we not have to treat it like a duck, I mean person?
Hmm, yes, I suppose that does show my bias - I'd really like to think there's something more than just chemistry going in my head. But then again, I've done enough science (and philosophy of mind) to see that it's very hard to argue otherwise.
Maybe I'll just have to resort to spurious linguistic arguments - should Materialists say that two distinct sets of bio-chemical reactions are better than one? Or that you accurately predicted the current chemical imbalance of my brain rather than you read my mind?
Materialism seems a very cold theory to my collective-noun-for-the-resultant-brain-state-of-interactions-between-bio-electric-impulses-and-bio-chemical-states-of-the-human-brain. :D
Good blog post, very interesting :)
Well, I'm a mathmatician and a philosopher (in training, at least), and so I can say that we can not say for sure that the brain is just a turing machine, since we don't know everything about it. Not knowing enough to say catagorically that "the brain is just a really complicated machine" is in no way going towards saying that cartesian substance dualism is in any way valid.
Try a religion, they're just about the only bunch of people with their heads far enough down into the sand to actually still believe in substance dualism :p
People who lack comprehension often think something similar.
I think people tend to think that because they don't understand something it goes beyond the reach of human ability no matter how advanced. Certainly if you showed a computer to a 16th centuary scholar they may still proclaim it as an act of unexplainable magic. After all (presuming you believe this) the human mind has evolved as a means of helping us survive which like all things in evolution are by no means perfect but if you follow the same logical build-up over time with a computer as a replacement you would eventually reach similar levels of sentience would you not?
Another issue is with the communications barrierand I'm sure it has been shown in some form that some computers may appear more intelligent because they follow the most human like functions as possible when in reality they aren't really very complicated at all.
Completely agreed. It is no less wonderful than any other explanation.
Definitely. In a strange way it kind of adds a more personal appraoch to running the site, and offers a closer relationship between the staff and the fans.
I really don't think there is a difference. Accurate prediction or measurement of the chemical balances in the brain, and the effects these have, would be reading your mind. They are not two seperate entities, which makes sense with the differences in approach between the two fields of thought at hand.
I would say so. I do believe we are simply computers acting out code, in the form of electrical impulses and chemical effects. So long as everything in your brain is copied with 100% accuracy, you would arrive with exactly the same brain and chemical make up. You would be exactly the same.
Great thread guys. Really enjoying it.
Enjoyment of philosophy is, IMHO, hampered by the fact the some philosophers were so far up their own ar$es it beggars belief.
The Rhino in the room assertion by Wittgenstein is a classic example, even Bertrand Russell thought he was acting like a fool.
http://www.amazon.com/Consciousness-Explained-Daniel-C-Dennett/dp/0316180661/ref=sr_1_2?ie=UTF8&s=books&qid=1241815955&sr=8-2
This book should be a required reading for anyone interested in AI.
Throughout recent history humanity has compared itself to a more complex version of the most recent technological development. In actual fact we have a better model for the human brain. It's called an artificial neural network. They can't do very complicated things yet - the most complicated task I've seen ANNs perform is learning to play chess (although I do have a dim memory of some research recreating the structure of an entire rat brain as an ANN, but I haven't read the paper, and it was in New Scientist which is sensationalist at the best of times, so that could be a vast exaggeration).
Heisenberg fail. ;)
It is still just a vastly more complex computer at the end of the day though, isn't it? There is still just an input of data, a process, and an output of data.
Well, be could start another blog post about the discussion it would raise, but I am a firm believer in determinism. Thus, unsurprisingly, I believe that anything that seems random simply appears that way due to our failure to comprehend or understand the methods required to predict and measure it correctly.
Agree to disagree, but I think there really is order to the universe. I am very much a "God does not play dice with the universe" kind of guy.
You're oversimplifying it in my book, but... OK. I stand by my previous statement though. I'd be more inclined to compare a brain to the Internet than to a single computer, but neither analogy fits as far as I'm concerned. And they ARE just analogies. I am fully prepared to accept the possibility that we will one day be able to map the exact structure of a specific brain and recreate it virtually (and just by-the-by, I don't think we'll have issues related to the uncertainty principle when we do), but just because we can reproduce the functionality of a brain within a computer, it doesn't make the computer a brain - there's lots of things we can simulate on a computer now, and we don't start arguing about whether the computer is any of them. I guess it's just too easy (and tempting) to anthropomorphize computers.
Order, yes. Determinism, no.
I think you'll struggle to find any modern physicists with you on that one, unfortunately. And as they're going to be the ones figuring out how to build your teleporter, you're going to want them on side.
Ha! You wouldn't want anything a physicist has made! Wait until the engineers have had a look at it. Physicists are lazy and usually don't solve any problems beyond zero or first order!
The teleporter would only work for a sphere of uniform density and mass (maybe even just a point object) teleported in a vacuum without the effects of gravity or magnetic field to interfere with it.