From: Tim Freeman (firstname.lastname@example.org)
Date: Wed Apr 02 2008 - 21:57:00 MDT
Rolf Nelson wrote:
> 2. Publish a technical document, or better yet a computer program,
> that would produce a Friendly outcome, given infinite computing
I think it can be done with finite computational resources. The
required resources for the relatively simple algorithm described below
are impractially large, of course. There's no practical difference,
so I'm nitpicking here.
> Presumably such a program would not significantly help
> someone to accidentally build a UFAI, but would provide evidence of
> whether the FAI community has reached some level of convergence as to
> the desired outcome.
From: "Eliezer S. Yudkowsky" <email@example.com>
>Actually, I suspect this is at least 95% of the way to a
>fully-functioning FAI, or UFAI.
Well, in that case the feco-ventilatory intersection is fast
approaching. My proposed solution is still out there, complete with
Python code that works for toy problems but can't run on real problems
and real hardware. The paper is at:
There's one known bug. It seems fixable:
Last time we spoke about this someone was demanding that we also have
a question-answerer, in addition to a Friendly AI. I sketched out how
to do that:
There hasn't been any significant progress since last time around,
though. Having a day job and kids keeps me busier than I would like.
-- Tim Freeman http://www.fungible.com firstname.lastname@example.org
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT