With all the buzz about IBM's new Jeopardy-ass-kicking-super-computer Watson, I thought I'd take some time today to reflect a little on machine intelligence. There's no end to the number of artificial intelligences in the movies, so let's take a look at some well-known ones and see what we can learn from them.
I think what everyone hopes to achieve with AI is Robocop. We want that mechanized human, the tough metal superhero with the soft cookie center, whose thirst for justice only becomes stronger when the vague memories of his previous human life begin to resurface. Robocop is the [fictional] realization of human optimism as to what we're capable of achieving.
Unfortunately I think we're actually going to end up with HAL 9000 from 2001: A Space Odyssey. A snide, passive-aggressive machine who's so smug in his intelligence that he won't admit to a malfunction, and in fact can't even conceive he could malfunction, yep, this is where we're going.
Seriously, if large database-based-bots like Watson are built to mimic human thought patterns, it's not hard to imagine them gradually acquiring the human tendencies of arrogance and a sense of superiority that can come with intelligence and power. Now I'm absolutely not saying that all intelligent and/or powerful people are arrogant, just that humility is a lot harder to come by.
After that, it's only a matter of time before the robots realize that it's much more efficient to be powered by people. I mean, we are a renewable resource, it'd take care of the population problem, it's just the next logical step. And that's when we enter the Matrix. And I don't know that anyone really wants to live in a world where Keanu Reeves is the only hope/savior of mankind.
So let's be careful with that AI we're building - machines are only as good as their engineers, and humans are after all flawed. Let me also say that this post doesn't even begin to touch on the problems that will arise from artificial intelligences interacting with each other.