Chapter Seven – Artificial Creativity

How do people create knowledge? David Deutsch’s simple answer is – we don’t yet know.

This chapter is about artificial intelligence, and the mostly misguided efforts to create it. Most people assume we’ve not yet created intelligent computers (artificial intelligence or AI) simply because our computers aren’t powerful enough yet. Deutsch points out how silly this argument is. If we had the program, even if it were too complex for the world’s most powerful computer, that program would be a good explanation. We wouldn’t even need to put it into a computer. Just by examining the program we could determine that it would work. At the moment, however, we have no such program.

Why? Because we don’t yet understand creativity.

As I wrote regarding chapter six, there are only two ways we know of that knowledge is created. The first is through Darwinian variation and selection. The second is through conjecture and criticism, the method used by people.

We understand Darwinian knowledge creation very well. Knowledge is instantiated in a living thing, a thing that can reproduce. Deutsch said, “The process of copying a genome is called a living organism.” (p 137) Through variation and selection, genes that are best at getting themselves copied into the next generation are the ones that predominate. So knowledge (genetic knowledge, generally a limited and parochial sort of knowledge, but knowledge nonetheless) is preserved.

But we don’t understand at all how humans produce knowledge. We know the key pieces, conjecture and criticism. But where does conjecture come from? This is mysterious. It’s no surprise, given the mystery of our own consciousness, that we can’t reproduce such consciousness even in computers that are probably well up to the task in terms of speed and memory capacity.

As Deutsch points out, we can’t fall back on the idea that because our computers aren’t powerful enough yet, it’s no wonder they can’t think. Computers will never think until we program them to. Deutsch has a simple rule about attempts to explain consciousness: “if you can’t program it, you haven’t understood it.” (p 146) The problem of consciousness and the problem of artificial intelligence are closely linked. Both lack a good explanation.