Hi Joshua,

Eliezer and I aired our disagreements in a long series of long series of
emails on the SL4 list several years ago.  As each of us understands the
other's views fairly well now, I don't think either of us has interest in
taking time to recapitulate the debate!

It's true that those old emails probably  make irritating reading, because
they contain a lot of boring yakking as well as useful content ... but
neither Eli nor I really has time to go back now and clean them up into a
beautiful literary dialogue ;-)

To very briefly address the "guaranteed Friendly AI" question that you
mention:
I do think that creating a provably Friendly AI (as Eli seeks to do) is an
incredibly hard problem, but I don't claim to know for sure that it's
impossible.  I think it's nifty that someone is working on this problem; but
I choose to focus my attention on the merely-extremely-difficult problem of
creating an AGI that looks extremely likely to be Friendly....  Eliezer
considers my approach too risky in terms of the odds of accidentally
creating a nasty AI; I consider his approach to have an overly high risk of
delaying the advent of Friendly AI so long that some other nasty danger
wrecks humanity before it happens...

-- Ben G

On 5/30/07, Joshua Fox <[EMAIL PROTECTED]> wrote:

> I have disagreements with [Eliezer] on many essential points

Ben,

I'd be fascinated to see your disagreements elucidated.  (I recall
reading about a difference of opinion is on whether guaranteed
Friendly AI is feasible. I'd like to see that explained in detail,
together with other points of disagreement)

If you've already written this up, perhaps I just missed the relevant
articles, so could you send a pointer.

Thanks,

Joshua

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?&;


-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=4007604&user_secret=7d7fb4d8

Reply via email to