Artificial intelligence

Michael Hartl — Sun, 02 Jul 2006 08:32:00 -0700

Just as the laws of physics allow the manipulation of matter on the finest scales, so too do they allow the possibility of machines that think. After all, (some) human beings think, and we are made of atoms just like everything else. The only relevant questions center on making a thinking machine—how hard will it be; how long will it take?

There are many scientists and philosophers who argue, for one reason or another, that artificial intelligence (AI) is impossible. What they usually mean is that it must be very hard, and they can’t imagine how to do it—that is, they imply that a problem of practice is a problem of principle. As in the case of nanotechnology, all of these arguments must ultimately fail:

Since brains follow the laws of physics, artificial intelligence is possible in principle.

(Of course, many people reject the premise of this argument: they believe that human beings transcend physical law in some way, perhaps through an immaterial “soul” or “spirit”. Such people may be right, but it is more likely that what we identify as “spirit” is information, which is still fundamentally physical in nature. In any case, there is currently no credible evidence that human beings violate known physical law.)

Artificial intelligence will have even farther-reaching consequences than nanotechnology (among other things, a sufficiently advanced AI would immediately invent nanotechnology if necessary), but the implications are harder to see—we’re simply too stupid to think of all the implications! Nevertheless, there has been much inspired speculation over the years; my favorite is Ray Kurzweil’s book The Age of Spiritual Machines.

Eikonoklastes by Michael Hartl: Artificial intelligence

Artificial intelligence