Re: Open Source Friendly AI? (was Singularity and thegeneralpublic)

From: Eliezer S. Yudkowsky (
Date: Fri Apr 13 2001 - 10:58:00 MDT

James Higgins wrote:
> I have found through many past mistakes that paranoia like this is a death
> sentence for your project. Because it greatly increases the chance that
> you won't succeed in the way you desire to.

One of the other factors that affected my decision was watching Eric
Drexler and the Foresight Institute, who went the path of sharing and
public awareness, getting systematically screwed over by scientific
community once the meme went mainstream.

Somewhat later I watched the excellent historical movie "Gandhi", which
reinforced that conclusion. The good guys can underestimate the
importance of control almost as severely as the bad guys overestimate it.
If Gandhi had spent a little less time working for independence and a
little more time worrying about what happened afterwards, there might not
be a "Pakistan" today.

Now of course, by saying this, I automatically sound like a bad guy,
because bad guys always talk about the need for maintaining control and
good guys always relinquish that in favor of fighting the immediate
battle. It's at times like this that I wish I'd started out with a more
neutral position, so I could be visibly moving towards the good side,
rather than having started with a too-extreme position (in this case,
open-sourcing the AI) so that I look sinister whenever I try to give my
reasons for moving back towards the middle.

-- -- -- -- --
Eliezer S. Yudkowsky
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT