TCS Daily

The Singularity Approacheth?

By Glenn Harlan Reynolds - October 12, 2005 12:00 AM

I, for one, welcome our new robot drivers. And they're well on their way, as the multiple finishers in the DARPA Challenge race demonstrate. Not long ago, robot driving seemed impossible. Now four vehicles "achieved a technological milestone by conquering steep drop-offs, obstacles and tunnels over a rugged 132-mile course without a single human command."

Are we making progress? I'd say so, since last year no vehicles finished, and the most successful only covered seven-and-a-half miles before failing.


At the risk of sounding Kurzweilian yet again, I think that this is significant, and in ways that go far beyond the matter at hand. It's progress, after all, and rather sudden and dramatic progress, at something that's rather difficult: Real world navigation. This is a development that goes beyond robot drivers. It has a lot of implications in robotics, as well as some bigger-still implications for the world as a while. It's a big deal.


It's also one that's getting some attention, but not all that much. As noted above, it's gotten mainstream news reports, and Popular Mechanics has certainly been all over the story, but -- much like the various items I mentioned in this column from a while back -- we've become so accustomed to rapid technological progress that we may notice the signposts in passing, but we tend to miss just how quickly they're zipping by.


Part of that is because we keep redefining progress. The Web, WiFi, and Google would have seemed incomprehensibly revolutionary not much more than a decade ago. Now we take them all for granted. They're just part of the furniture.


Likewise, just as chess-playing machines that could beat human grandmasters were once the very definition of artificial intelligence, now they're just trivial accomplishments: toys, really. And, I suppose, within a few years self-driving cars won't count as "real" robotics either -- just helpful gadgets. Yet as the signposts flash by, we get closer to the destination, whether we notice it or not.


One of the big criticisms of futurists who write about the Singularity is that the kind of strong artificial intelligence that Singularity-enthusiasts predict we'll accomplish is hard to achieve, and hasn't been achieved yet, despite the extravagant promises of researchers decades ago who thought that they could achieve human-like intelligence on machines less powerful than a Gameboy. (Much less powerful -- though to be fair, they did fill a room much more impressively). That's certainly true, and we're still a long way, in terms of capabilities, from HAL. (Or even -- despite the success of the DARPA racers -- from Keith Laumer's self-aware Bolo combat robots.)


But we were a long way, in terms of capabilities, from self-navigating cars not long ago, too. And being a long way from something in terms of capabilities isn't necessarily the same as being a long way from something in terms of time, when your capabilities are improving at an ever-increasing rate. The faster those signposts flash by, the less time it takes to reach your goal, however far away it is.


TCS Daily Archives