All Hail Our New Robot Overlords

Lots of folks are talking about Ray Kurzweil's new book The Singularity is Near. His argument, basically, is that true artificial intelligence is a function of computing power, we currently haven't created it because we don't have the computer power, given current trends we will have it in about 20 years, then our artificially intelligent robots will begin working on speeding up the process ever-more, making human intelligence almost useless in a relatively short period of time. Kurzweil kindly goes in for the "this will help humans and make us all much happier" explanation rather than the "we're all gonna be robot-slaves" argument. To me, that one seems a coin toss. In any case, the singularity is when it happens, when our intelligence becomes increasingly non-biological and and the world becomes Totally Awesome.

Reactions vary. Kevin thinks he's right, but that he cheated on a graph. Matt thinks he's wrong, and points to our dashed hopes for nuclear power as proof. Tyler wonders why IE still crashes if we're so damn smart. And so on.

Count me critical. Inventions don't tend to follow the tracks we think they will. If they did, we'd have long ago had our flying cars, phaser guns, and teleportation devices. So the graph that Kevin references showing the increasing speed of technological innovation strikes me as a point against. We're getting better at making things, but we don't tend to improve them in the way futurists expect. Very rarely folks get a few things right and so the reputation of futurists everywhere is saved, but given what we've seen in the past, Kurzweil's analysis strikes me as too easy an extrapolation to accurately describe where we're going to end up. In my read, the one thing we aren't is linear.

On that note, what sort of intelligence is Kurzweil talking about? I assume he's after something quite similar to consciousness, which may prove a problem. Kevin may be right that the brain is but a biological machine, but it's a biological machine that doesn't quite follow itself. Many of its properties are emergent, and unless we're starting to see that correct programming and processor design are giving rise to characteristics much different than producers expected and more powerful than anything they imagined, AI is going to probably remain quite a ways back in consciousness, simply because we've little idea how to actually design anything actually resembling it.

Now, if Kurzweil is talking about glorified calculators, massively able machines that do a couple things a billion times better than humans and are considerably more flexible when working on a problem, that's likely possible, but nevertheless a whole different, and potentially less arresting, issue.

Lastly, and this is decidedly not what Kurzweil's talking about, we have a society raised on Aasimov and The Matrix -- the barriers to AI, in the long-run, are going to be less technological and more legal/societal. When truly smart machines do start to tumble off the production line, when they do begin to replace folks in service jobs and prove able to replace them in all jobs, you're likely to see a backlash of almost unimaginable proportions. Nativism will be supplanted by its far more desperate cousin, humanism, and I'd be fairly surprised if the government didn't enact controls making the prohibitions on cloning look laughable. Think the X-Men comics in more recent years. Humans don;t like to believe themselves obsolete, and my hunch is they'll be vicious and swift in stopping production on anything that threatens to make them so.