Friday, March 21, 2008


Luckily, I'm not the only person blogging from ICFA. Cheryl Morgan is going to lots of panels that I'm not, so between the two of us we're covering pretty good ground.

Another panel I'd like to mention is "Politics and the Singularity" from last night. Membership: John Fast (moderator), Vernor Vinge, Joe Haldeman, Robert J. Sawyer, James Patrick Kelly, Cristopher Hollingsworth.

Like so many panels about the Singularity and politics, this one never really got around to its subject. It was pulled off course by discussions of how to define the Singularity, all the different kinds of Singularities there are, and what their probabilities might be. At the end, I still declare myself to be a Singularity skeptic (although I do quite enjoy reading fiction about it).

However, a couple interesting points were made that I hadn't heard before. There was discussion of an arms race to build a Singularity-starting super AI computer, with two sides racing to be able to turn it on first. Sawyer pointed out that really, the side to "push the button" first would lose: they'd sacrifice all their power to the Super AI, which would presumably make all their decisions for them, and effectively enslave them. The problems I have with this scenario are the same ones I have with the Singularity concept altogether (it's unlikely people will, or will be able to, design AIs like that), so there may be some interesting dramatic potential there. Fast, the moderator, pointed out that politicians are likely to think of this sort of thing as a magic genie, and probably would push that button without fully understanding the consequences.

Haldeman put forth the theory that the Singularity might take a very long time, on the order of centuries, and in that case we're probably already in the midst of it. In the long run, that makes more sense to me than an over-night take-off. Also, I finally found out what IA means as opposed to AI: Intelligence Augmentation, i.e. designing better humans and better human intelligence. I can more easily see that leading to a Singularity-type scenario than AI.

All in all while I found some things disappointing in that panel, when you've got people that smart and interesting talking, you'll always find something new to think about.

No comments: