Michael Anissimov: Luddite

By | June 9, 2009

That’s right, it’s a throwdown. I’m calling him out. He may be one of my favorite people, but what else am I going to say when he writes something like this?

Another product called “Tesla”, Nvidia is selling supercomputers up to 250 times faster than standard PCs and workstations for just $10,000.

I’d prefer if this sort of product weren’t around.

See? Luddite. What else can I say when he approvingly publishes a quote that ends with this line?

Moore’s Law is the enemy.

Wow, what a drag. Maybe Michael and Leon Kass can start vacationing together or something. Sheesh. Sorry to have to take such a hard line, but what else can I say?

Well, okay, there’s one other thing I can say: Michael makes an excellent point. And, of course, he’s not a Luddite any more than he ever was. He’s merely an observant student of Moore’s Law for Mad Scientists.

To tell you the truth, I’m torn. I mean, on the one hand, we need for supercomputer capability to become commonplace if we’re going to get to a world where anybody who wants to can pretty much run his or her own space program. But we’ll never get there if, along the way, somebody using that same ubiquitous computing technology accidentally builds V’ger and Kirk and friends aren’t around to talk it out of wiping out all us “carbon based units.”

And it isn’t just the accidental stuff we need to be concerned about. As our need for virus protection software indicates, there are already plenty of folks out there in computerland with agressive and hostile agendas. Making super-fast, super-powerful computers more widely available can only empower such individuals.

On the other — wait, what am I on, like my third hand, now? — a rising tide raises all boats. Widespread supercomputers also enable careful AI reserachers and virus-fighters. Maybe the availability of the Nvidia is just part and parcel of a world in which friendly AI can be achieved.

Anyhow, here’s hoping.

  • Jonathan

    I agree that we need to work on the Friendly AGI problem, but bitching and griping about faster computers isn’t very constructive.

  • Nick Tarleton

    Widespread supercomputers also enable careful AI reserachers

    Not really; friendly AI is overwhelmingly a philosophical/mathematical/logical problem. As a rough analogy, the ability to produce bigger explosions doesn’t, above a certain very low threshold, make it easier to safely demolish a building. There are probably world-savingly useful things that very careful people could do with huge computing power, but (given wide distribution) these are probably completely outweighed by the danger of uncareful AI.

  • Tom

    Third hand = Gripping Hand