The Ghost in the Machine
Captain Futuro Predicts
A lot of debates about artificial intelligence come down to "what about consciousness? We don't understand consciousness."
I will stipulate to that. Consciousness is the last refuge of the doubters - the last hiding place for some kind of magical hocus-pocus - unless you count quantum gravity, or like Roger Penrose, conflate them. Well maybe, but I doubt it.
The fundamental building block of consciousness is self-awareness. Our consciousness is our awareness of self, and our awareness of other's consciousness. Old style machines had no self-awareness. A truck was just a truck, not knowing or caring about other trucks or even itself.
It's pretty clear that other animals, even rather primitive ones, have some level of self-awareness. It's an essential tool for competitive success in the struggle for existence. Our trucks and other machines are developing some self-awareness too. They have computers which keep track of when they are low on oil or developing certain types of mechanical trouble. Some of them will keep track of where they are, and some will announce to their drivers when they are not feeling up to par.
This is not to say that your pickup is going to start arguing with your taste in music soon, but the seed is there. Designers of so-called autonomous vehicles are well aware that a quasi-biological self-awareness will be crucial for their effectiveness.
There is more to consciousness than keeping track of your oil pressure, but how much more? Well, you need a model of other's thinking processes. This is crucial for most of us, but autistic people seem to make do without it. A lot of the information we get about what others are thinking comes from such things as facial expressions and tone of voice. There is no doubt that computers with suitable sensors can do that very well already. Soon your door may greet you with "Uh, oh! Looks like *you* had a bad day."
I think this future is coming, and coming fast. It's likely that machines won't spend much effort understanding how we think. They will probably be a lot more interested in how other machines are thinking.
I think I would prefer to be wrong on this.
Comments
Post a Comment