From: Ben Goertzel (firstname.lastname@example.org)
Date: Thu May 02 2002 - 21:52:58 MDT
> At the recent Foresight Senior Associates gathering (April 26-28, 2002,
> Hotel Crowne Plaza Cabana, Palo Alto) , the Singularity was featured in
> several presentations (although not mentioned in the website program
> preview: http://www.foresight.org/SrAssoc/spring2002/program.html ).
> Foresight noted: `For the first time, this meeting is not
> "off-the-record."' I know Eliezer and John Smart spoke; perhaps others in
> the SL4 forum also attended. As a distant but interested Australian, I'd
> welcome any news reports of this potentially important occasion.
> Damien Broderick
I was there, as a representative of the Frankfurter Allgemaine Zeitung, a
German paper that I write for sometimes.
Frankly, there wasn't much said on the formal program there that would be
novel in any way to folks on this list. I think that the Foresight Senior
Associates Gathering exists mainly to bring Silicon Valley investors and
businessfolk up to speed on radical techno-futurist thinking. There are no
highly technical presentations; it's more a matter of techies and policy
wonks bringing the investment/business audience up to speed on recent
developments and ideas.
That's not to say there weren't plenty of interesting conversations "off the
For instance, I saw a demo of Peter Voss's Artificial-General-Intelligence
in the making, a highly sophisticated, vaguely neural-nettish program that
is learning to recognize patterns in visual scenes and other things, and to
carry out actions coordinated with its perceptions etc. Still in the early
stages of development, but definitely interesting & impressive.
(Of course, Eliezer, Peter and I each have our skepticism about the others'
approaches to AI; but even though Peter's approach is not my chosen
approach, I still think it makes good sense and has a nonzero chance of
success at the grand goal. I think there are *many* routes to real AI,
although I do think my route is the best one I've seen so far ... and I know
Peter thinks the same about his ;)
And Steve Jurvetson (the VC from DFJ), as well as walking around with a
silly-looking blinkie in his ear, was pursuing a long conversation with
Eliezer and others on the question of whether evolution *necessarily* leads
to selfish creatures, or whether it would be pragmatically possible to
evolve creatures with compassion as the fitness function.
There was one presentation that touched on notions of Friendly AI, but
didn't directly refer to Eliezer's work, and in my view presented a somewhat
simplistic perspective. It was proposed that "greater intelligence
automatically leads to greater morality because it allows beings to better
see the consequences of their actions." I'd *like* to believe this was true
in some sort of statistical way, in spite of numerous obvious anecdotal
counterexamples, but no convincing argument was presented. This
presentation was a little disappointing in that one would at least have
liked it to REFER to Eliezer's presentation from last year, so as to give
the feeling of incremental progress from year to year. But I suppose that
philosophy rarely gives the same feeling of ongoing progress in the way that
science and engineering do.
Ray Kurzweil gave a nice "special interest group" on the Singularity. He
made pretty clear, in response to some pointed questions by Peter Voss after
the formal meeting was done, that he believes the "imitate the human brain
closely" approach is by far the most likely path to artificial general
intelligence. I think this is close to the view of Eugene Leitl, on this
list, and others as well I presume. After enough repeated exposures I have
to say I get a little weary of Kurzweil's graphs of the impending
Singularity -- but even so, I'm glad he's out there tirelessly making the
point, which the vast majority of the educated human race still refuses to
On the more practical side, Barry Bunin demonstrated some really cool
biochemical database software being made at his company Libraria. This kind
of work -- providing databases of knowledge spanning multiple areas of
science -- is paving the way for massive scientific progress in a kind of
low-key way. 'Cause it's these databases that near-future AI's will study
to figure out how to improve humanity genomically...
A change from last year is that this year there were a few venture
capitalists *eager* to talk to entrepreneurs. Last year on the other hand,
the investor community seemed to be in a totally quiescent phase. Now it
seems almost half-awake ;-p
Etc. There was plenty more, this was just a quasi-random sampling...
-- Ben G
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:38 MDT