Artificial Intelligence
I’m going to write a series of posts on artificial intelligence. I don’t know how many posts there will be, I’m just going to keep writing about it until I’m done. I have many thoughts and opinions on the subject, as I’ve been thinking about it off and on for about 14 years now. But I make no claims to originality, since I have not taken any formal course in AI nor have I made any rigorous study of the scientific literature. It’s entirely possible, in fact, it’s quite likely that I have not come up with a single original idea in 14 years. I just wanted to see how far I could figure out things on my own, under my own power, without standing on the shoulders of anyone, giants or otherwise.
In part, my refusal to study AI formally comes from a peculiar combination of intellectual laziness and conceit: laziness, because it takes a real effort to learn to think like other people do in order to follow their arguments and theories, and conceit, which manifests itself as a fear that, having learned to think like other people do and see the problems of AI from their perspective, their trains of thought will completely overwhelm mine. But what reason do I have to believe that my thoughts and ideas are at all superior?
I have picked up a few things here and there, however, and will try to cite sources whenever possible.
So. First question is, what is artificial intelligence? The short answer is, nobody knows, or, to be more precise, nobody can agree on what the name should mean. It’s a very large field, and not everyone is particularly interested in the same problems. Since different problems lend themselves to different solutions, and since everyone wants a slice of the AI pie, it should come as no surprise that the field is so fractured. I am going to adopt a very, very general definition, explain it a bit, and then in my next post I’ll take a look at different AI sub-fields, with my thoughts on each.
My definition of an artificially intelligent program is any program which reacts to its environment in an organized, intelligent way, usually by attempting to mimic or perform some function of a living, thinking organism. Now that’s pretty broad, everything from toys like Furbies and Tamagotchis to chatterbots to video game characters to robots which can think for themselves fall under this umbrella. But it’s a necessary definition. Programs which can help doctors diagnose heart problems and programs which automatically translate text are both artificially intelligent, to some extent, but they are not interchangable! It’s kind of like defining the word “math”, there’s many many different kinds of math, and the kind you use depends on the problem you want to solve.