Rod Brooks spoke most interestingly at our join BCS/RSI lecture in 2007 on robotics. As an aside, this was one of the drivers behind an event with the Royal United Services Institute on autonomous weapons which I will no doubt say more on very soon.
What has inspired me to engage in my own literary equivalent to the winter vomiting bug is what Rod Brooks said to The Edge Foundation he had changed his mind about over the last year. His answer, titled 'Computation as the Ultimate Metaphor' is worth a read. Here's a quote:
'The metaphors we have used in the past for the brain have not stood the test of time. I doubt that our current metaphor of the brain as a network of computers doing computations is going to stand for all eternity either.'
If I were trying to be sensationalist, I would say that this is tantamount to AI heresy from the Director of MIT's computer science and AI lab. Shocking!
However, I'm not, and it is a pretty sensible point. He demonstrates how easy it can be to start confusing the model (or metaphor) with reality.
'Some of my colleagues have managed to recast Pluto's orbital behaviour as the body itself carrying out computations on forces that apply to it. I think we are perhaps better off using Newtonian mechanics (with a little Einstein thrown in) to understand and predict the orbits of planets and others. It is so much simpler.'
Being simpler doesn't just mean easier to use, but easier to forget its a model rather than a reality. The phrase 'it breaks the laws of physics' is a happy fiction, in that something can break our laws of physics, but can't break physics. Physics is de facto what happens, so anything that happens can't be against what happened, because it just did. Simple.
If only it were that simple. Quantum stuff doesn't have to happen in order to have an effect, it just has to have the possibility of happening. This is why mathematically-based computation hits a wall, and why tachyons may be able to interfere with causality even if they don't exist.
Take the classic problem - build the perfect chess computer. One thing that computers are very good at is examining lots of possibilities at once - like brute force attacks against a code. The easy solution is examining all the possible moves in the chess game from a given position, and working out which produces the best chance of a successful outcome. It is how we idly imagine chess masters approach their games.
The only problem is that the number of possible chess moves in a game is so large as to have its own name - the Shannon number. This is somewhere in the region of 10120 which is really quite a big number. To put it in context, this is approximately 1040 larger than the estimated total number of atoms in the universe. That presents a problem. Even if you were able to use each atom to store multiple bits of data, and use every atom in the universe to do so, you'd still need more universes than we currently have available.
Wantonly using the sort of facile pop-science logic that may get me beaten up later, I would suggest that this means a simple chess board contains more possibility than the universe is able to computationally deal with. The algorithm can be described, just not followed to its iterative conclusion. Quantum computers are better able to deal with problems like that (which is why they may be rather good at cryptanalysis).
From the point of view of discrete mathematics, or the Church-Turing Thesis, quantum computing has to be marked as 'cheating'. It is making use of physical processes that are non-computational in order to solve a problem. Is that then computation? (I don't know, I'm really just bluffing my way through this today)
There are those that believe that the brain is computationally-based, just extremely complex (the phrase "what else could it be?", when used, shows a worrying lack of imagination). There are those that believe the brain is based on computational and physical interactions in the same way as a quantum computer. There are those that believe it is something else (either holding very strong views on what that something is, or none whatsoever). Finally, there are those that don't care, as long as it works.
Ironically, and probably of comfort to those of you who have reached this point in the blog and realised you don't care either, the last view is perhaps the closest to a purely scientific viewpoint. Belief is of course, to use a horribly post-modern word, 'valid', but science is ultimately empirically based (sorry, mathematicians). If you're starting to believe that physical models are really what happens, that F really equals ma, that E really is MC2, then its probably time to unplug.
