[extropy-chat] singularity conference at stanford

[John K Clark]

"Eugen Leitl" <eugen at leitl.org>

> I see no continual improvement in AI which will lead to robust, natural
> intelligence.

Years ago I remember being amazed that a home computer could search through
a very long document for a particular word or phrase, it seemed like magic;
if at that time you'd told me what Google could do with the sum total of
human knowledge (or nearly so) I'd have said it was imposable unless Google
had real intelligence. And I seem to remember that Hofstader once said that
a computer could never beat a chess grandmaster unless the machine was
intelligent and had a profound understanding of the game. However Hofstader
has changed his mind about that and that's the problem, true AI seems to be
whatever computers can't do yet, and by that definition we will never have
true AI.

> The bad thing about discontinuities is that they're so hard to predict.

That's true but I don't think you'd need a discontinuity, like a quantum
computer (although that would be very nice to have) to make a AI, just keep
getting twice as good every 18 months and you'll get there.

If I were a singularity skeptic (and I'm not) I wouldn't concentrate on
computers but on medicine. It's astonishing and discouraging how an enormous
increase in biological knowledge hasn't translated into cures. There was a
discontinuous improvement in medicine in about the year 1900 with the
invention of anesthesia and sanitation, for the first time in history
medicine did more good than harm. There was another discontinuous
improvement in medicine in about 1950 with antibiotics, but after that
nothing dramatic, it's just been busy work. A very few people with rare
diseases are a lot better off then they would have been in 1950, but most
people with common diseases just take a few extra painful weeks to die.

And Eugen, you write some good stuff, so I sure wish you'd get a better mail
program, it's a pain to open a attachment every time I want to read what you
say.

  John K Clark