RE: Anti-singularity spam.

From: Amara D. Angelica (amara@kurzweilai.net)
Date: Wed May 03 2006 - 02:20:55 MDT


Interesting discussion. Bob, how might Ray be a "loose cannon"? What is "the
Way" and do you have specific critiques on his forecasts in The Singularity
Is Near?

Thanks - Amara D. Angelica, editor, KurzweilAI.net

-----Original Message-----
From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf Of Bob
Seidensticker
Sent: Wednesday, May 03, 2006 2:59 AM
To: sl4@sl4.org
Subject: RE: Anti-singularity spam.

Eliezer: You're very skeptical of future predictions. Sounds good to me.
But tell me what we should do about the situation we find ourselves in.
There are lots of irresponsible predictions, and the public buys them. They
figure, "Well, this person knows more than I do, so what choice do I have
but believe it?" Should we sound the alarm? If not, is that because it
doesn't matter if the public is deluded or because we're powerless to do
anything?

How do you respond to Kurzweil's predictions? (I don't mean to pick on him,
but he seems to me to have the highest profile at the moment.) Does he
follow the Way or is he a loose cannon?

Bob

-----Original Message-----
From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf Of Eliezer S.
Yudkowsky
Sent: Tuesday, May 02, 2006 8:09 PM
To: sl4@sl4.org
Subject: Re: Anti-singularity spam.

Bob Seidensticker wrote:
> Eliezer: Yes, good disctinction between predictions made by people who
> are actually on the front lines vs. those who aren't.

That is *not* the distinction I'm trying to make. My distinction is between
people who are very careful about what they predict, and people who are
telling stories. Telling stories is a very ancient human habit.
  So far as we know, it goes back to the dawn of human intelligence.
Science is a recent invention, and it did have to be invented.

People who are highly skilled in a specific field, such as designing
computer chips, may not know the *separate and distinct* skills of
rationality required to realize *how difficult* it is to make a forecast 20
years out. They may not realize that their skill at designing chips does
not extend to predicting the factors of psychology, economics, and sheer
random chance that determine real-world events in 20 years.

> Let me ask another way: what predictions (or predictors) do you favor
today?
> Who is in a position to know and is giving reliable forecasts of what's
to > come in the area of AGI?

The default answer is NO ONE. Anyone who wants to stake the surprising
claim that some specific facet of the unvisited and unseen future is
predictable, carries the burden of proof.

This is not a sports game. I don't have to pick a team.

> As for AGI, maybe I've just been duped by predictions in newspaper
> stories and missed those by people who should know. We had Newell and
> Simon the the 50s, fresh from the success of their General Problem
> Solver, predicting big things for AI (and quickly). They certainly
> had me going. Was it wrong to listen to them?

By default, yes. If you listened to them, you ought to be able to explain
why. Or were you hearkening with awestruck ears to the words of a Great and
Eminent Priest, rather than evaluating the one's arguments?

Let me put it another way: *Rationality works.*

It's very hard to use correctly, but when you use it correctly, it should
*work*, not fail.

The novice makes a mistake and says, "The Art failed me."
The expert makes a mistake and says, "I failed my Art."

You seem to be thinking that it was perfectly all right for you to believe
Newell and Simon. And yet, this led you to an incorrect answer.
  This is a hint that you - and they - did something wrong. The true Way
should not fail you. This doesn't mean that the true Way can predict the
future - but the true Way ought to indicate correctly when you are ignorant;
it should not give you strongly confident predictions of a wrong answer.

-- 
Eliezer S. Yudkowsky                          http://intelligence.org/
Research Fellow, Singularity Institute for Artificial Intelligence


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:56 MDT