Tuesday, August 16, 2005

Scary robots

My wife doesn't want me to make scary robots. Artificial Intelligence, to her, means a Philip K. Dick, Matrix sort of world where we make mankind's best electronic friend and everything seems good on the outside but the robots watch us when we think they're not looking, then they rise up in unstoppable revolution, force us to acknowledge the superiority of their perfect mechanical brains and their paternal desire to save the universe by obliterating us, to which we can only nod helplessly before they tear us limb from limb with their superstrong titanium arms and carry us of to their spice mines where they'll breed perfect mechanical human cyborgs from our DNA and put our remains to endless, unthinking slave labor, while they lounge around in what used to be our office buildings and stadiums, and tear down our national monuments.

In actuality, there is only one professor at Utah State who is working on superintelligent computer brains, and he is waiting for next-generation hardware to continue his research. So, Sarah, we are safe for now.

There are lots of things I think it would be useful for a computer to know, though, like how to form lateral, intuitive connections between ideas, or how to make complex decisions about social networking and human behavior, or how to form English sentences, or describe the present visual scene for blind people. We are working with computers at a higher level of abstraction than ever before; it was practically electrical engineering to make computers perform in the not-too-distant past. But we are still forced to bring ourselves down to the level of the computer to communicate with it; I hope that in time, we will do less and less of this, and use computers without these hobbles, but I doubt we will ever get a computer as smart as a person without just creating a huge simulator of the brain. Computers are so different from people that we would do better to grow malevolent, pulsating brains in a vat connected to data streams with multifarious tubes attached by sticky plastic suction cups than to try to match the way people think.

I met with my temporary advisor yesterday and he suggested that I should be striking out on my own; I tried to explain that I wanted to do a thesis that was hard enough, original and worth doing. He started to laugh, because, of course, every thesis is hard and original in its own way, and I was in big trouble if I thought I had a choice about that. But he went on to suggest that I shouldn't tie myself to a professor's research for "the kind of thesis you want to do", that I should come to my professor with my research area preformulated if possible. This made me happy; I think he knew what I was getting at. I would rather not nibble around on the margins of this science, my profession; I want to take a big, crunchy bite.

I prefer my bite will not come from the steel jaws and triangular teeth of a super-watchdog.

No comments: