If it's in a computer, can it be real?

My brother got me to go to a talk/discussion session with Steve Grand (of Creatures fame) last night and it turned out to be very good. The talk was part of the AV Festival, taking place across Newcastle/Gateshead (they really are slowly turning into the one city), Sunderland and Middlesbrough and was entitled 'Artificial Life and Virtual Reality'. The base point of discussion was whether anything generated in a computer can ever be 'real', with a focus on intelligence/consciousness in a computer. Considering the talk and following discussion spanned two hours and isn't exactly light going, it's quite hard to summarise, but basically Steve Grand says that a computer program can never be written to have a 'real' intelligence/consciousness/whatever you want to call it, but that a computer can (theoretically, with a powerful enough computer some time in the future) simulate an environment in which a 'real' thing can emerge: if you can model and simulate every atom in a person's body within a computer, with the basic rules that apply to those atoms, then you should effectively have another version of that person, as the interactions of the atoms combine to create that person, their memories, etc. I no doubt explain it particularly poorly, but then that's why I'm not off being paid to give talks or write books on the subject. I do have a bit of an interest in artificial intelligence from working at a computer game studio and then writing my degree dissertation on path-finding for computer games though, and I have to agree with Steve Grand that all 'AI' work so far is not in fact intelligent, it's just giving the impression of intelligence through a series of tests. That's pretty obvious - especially within the computer game world, where not much processing power can be handed over to the AI - but it's got me to thinking about this theoretical point in the future when we might be able to have 'real' intelligence emerge through the creation of a suitably complex simulation of atoms/particles. If we could create real intelligence within a simulated world, what limits should we then put on its application? Once we got to the point where we could generate a game world suitably complex that we could insert copies of people that could think independently, it would open a debate over whether they could continue to be used in the traditional way for computer game characters - often to be shot at or otherwise killed. If they have real intelligence, they surely have real feelings and so should not be subjected to being shot at for the entertainment of some 'higher beings'. The argument could be there that in the end they're still just lots of ones and zeros flying around in a computer, but as Steve grand pointed out last night, in the end so are we, just replacing ones and zeros with protons and electrons. Luckily I don't think it's a problem we'll have to contend with too soon (though you never know). Anyway, Steve's talk and his answers to a very wide range of questions afterwards were great, so kudos to him and kudos to my brother Christian for making me go see someone I'd never really heard of ;)