This is almost two years old, so they may be further now. On the other hand, from what I read in the book Physics of the Future, ASIMO is not as advanced as it seems in demos. Everything is carefully scripted, and it can't actually perceive its surroundings, unless they are specially marked. It also does not think or generate its own speech, it just plays recordings.
But the movements are better than anything I've seen.
That will probably be true of any artificial intelligence. The kind we see in some science fiction, like Iain Banks' Culture, which is self aware and as alive as we are, probably isn't possible. I'm not a programmer but it seems like no matter how sophisticated it might be an A.I. can only react the way it's been programmed to react and in the future this might result in A.I. so sophisticated it can perfectly mimic sentience but it won't be sentient.
ReplyDelete
ReplyDeleteyes, and that makes for very interesting questions. How do we judge sentience? If a robot acts just like a human, can we be sure it's not sentient?
And why would something make of meat be sentient while silicone wouldn't? How'd that happen? Is there a "soul" different from the body which is the sentient part?
If a robot acts just like a human, can we be sure it's not sentient?
ReplyDeleteBecause if it's programmed to mimic sentience it can't have real sentience.
And why would something make of meat be sentient while silicone wouldn't?
The difference is that in any biological organism you have cells which by themselves achieve nothing but come together to produce something else - blood, muscle, bone, brain cells. Consciousness is the result of billions upon billions of cells which individually are nothing but working together produce consciousness.
With artificial intelligence, it's a program, it's electricity running through circuits. No matter how sophisticated that becomes, it's still just a program - a bunch of if, then statements and stuff like that. Biological organisms aren't programmed in the same way - the result of greater adapatability. Put another way, you could argue that individual cells are programmed but what they together combine to produce isn't programmed. The adaptability is the key. A machine can only react in ways it's been programmed to react.
As Michio Kaku said in Physics of the Future, brute computational power does not equal intelligence. Deep Blue beat Kasparov (and you can get cheap chess programs that will beat anyone not a world champion) but it can't actually think - it can only play chess. While it's tempting to say "Well, yes, but in the future--" and maybe, but most likely computers' brute calculating power will vastly increase but they won't ever be able to actually think.
ReplyDeleteYes, that's an important point. Many scientists seems to think that computational complexity equals intelligence.
ReplyDeleteBut some things are missing: Self awareness, importantly. And the ability and the desire to *originate* thoughts and actions.