Re: How hard a Singularity?

From: Eugen Leitl (
Date: Sun Jun 23 2002 - 04:18:06 MDT

On Sat, 22 Jun 2002, Mike & Donna Deering wrote:

> James, when we say human equivalent we don't mean performance, we mean
> algorithms. Doing the thinking using the same algorithms that a human

You don't make much sense here, I'm afraid.

> uses. We assume we can figure out these algorithms because we have an

Can you please write these algorithms down? Where is the tarball I can
pick up that utilizes these algorithms to achieve something constructive
(like classifying online images into violence/pornography/safe for
consumption by small furry mammals bins)?

> example to study. These human equivalent algorithms will run faster
> and more effectively on a silicon substrate than a biological one.

Wishful thinking here, pure and simple. And please don't mention silicon,
silicon is about to go away way before we have AI. (Good riddance). If you
think you can map full-detail neural emulation to a chunk of computronium
efficiently, good luck. You're going to need it.

You realize that your arguments so far amount largely to a lot of
handwaving? I'm not disagreeing with you in principle (feasibility of
cognition on an artifical substrate), but you have to make a much stronger
case than just a lot of unbased assertions.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:39 MDT