Eliezer Yudkowsky - Introducing the "Singularity"

[Heraclitus]

[Al B. Sure!]

What happened to the 2 new topic per 24 hour rule?

Your threads are sucking, Hera. You've gone from walls of text to crappy lectures.

And why is he reading from the computer? God people have no public speaking skills

If he knows this stuff, he should be able to do it impromptu.

[Lorizael]

I've read a lot of his stuff. What's he saying here?

[Heraclitus]

I've read a lot of his stuff. What's he saying here?

He talks about several different technological singulatarian concepts (accelerating change, intelligence explosion, event horizon, ect.) and the differences between them. If you read any of his stuff and hang around that particular memosphere you probably won't hear anything new.

[Heraclitus]

Your threads are sucking, Hera. You've gone from walls of text to crappy lectures.

Are you sure you want me to return to ranting walls of text?

[Al B. Sure!]

Michio Kaku shows how public speaking is done

[Lorizael]

Michio Kaku shows how public speaking is done

Maybe, but he's also kind of a loon.

[Al B. Sure!]

Maybe, but he's also kind of a loon.

A clever and engaging loon who knows how to rock the audience!

[Al B. Sure!]

Oh and his books are very accessible. Hyperspace Parallel Worlds

[Ari Rahikkala]

I'm not even sure why people bother using the word "singularity" for anything but the Vingean kind of thing. It's the only one that really says "and here's the point where rules that previously worked break down and predictions of the future become useless".

All Kurzweil is arguing for is that technological development (yes, techno-juju) will keep growing exponentially for at least the next several decades. That's what we would, in mathematical terms, call... not a singularity. If only there were a sexy name for it .

Similarly, all Yudkowsky is arguing for is that if you make a system that has its own intelligence as part of a recursive self-improvement loop, it's likely to surpass human intelligence and go far enough beyond it to be useful in solving difficult problems. This idea actually even has a sexy name: Intelligence explosion. There's no reason to muddy up a perfectly good engineering problem like that (the problem of actually implementing such a system and doing it safely) with some sort of mystical techno-millenarianism or poorly founded extrapolation.

Really, this guy hasn't been spending too much time talking about any predicted intelligence explosion anyway. He's too busy trying to make the world a halfway sane place before he can really get started . If you're going to read Yudkowsky, you might as well check out Three Worlds Collide, a story involving things like aliens that greet humans with tentacle porn, and spaceships where major decisions are made by prediction markets. It was written, quoth the author, "to illustrate some points on naturalistic metaethics and diverse other issues of rational conduct".

[Heraclitus]

Like I've said before I've read quite a bit of Yukdowsky's work, I actually started with the Three Worlds Collide story.

[Heraclitus]

[KrazyHorse]

I especially like the last sentence:

"Due to the minimal effort you put into friendliness, human life exists in less than one out of every hundred billion quantum branches in which you created an artificial general intelligence; in the other branches the artificial general intelligence is eating everything in its light cone to maximize its control of free energy"



EDIT: replaced "that" by "than"

[Kuciwalker]

what?

[DriXnaK]

MAD TV - Introducing the "iRack"