On Fri, Aug 22, 2008 at 3:23 PM, Valentina Poletti <[EMAIL PROTECTED]> wrote:
> Thanks Vlad, I read all that stuff plus other Eliezer papers. They don't
> answer my question: I am asking what is the use of a non-embodied AGI, given
> it would necessarily have a different goal system from that of humans, I'm
> not asking how to make any AGI friendly - that is extremely difficult.
>

If AGI you are talking about is not going to be powerful enough, it is
a completely different question. Weak optimization processes are the
stuff we build our instrumental actions on, by creating contexts from
which these processes can't (aren't supposed to) break out, and so
exercise their local optimization in a way that is bound to lead to a
different end.

I don't see what relevance does the choice of embodiment has to
anything, except for practical considerations during the earliest
stages of development, when AGI is not yet able to model sufficiently
high-level events in the environment. In today blog post, I concluded
the long arc of posts that describes a high-level perspective on
operation of AGI agent ("holistic control"), that emphasizes how a
particular way in which input and output are organized (particular
embodiment) is insignificant. A quote (
http://causalityrelay.wordpress.com/2008/08/22/holistic-control/ ):

> The operation of control algorithm is focused on the support of model of
> environment, not on action and perception. Action and perception are only
> peripheral (although indispensable) aspects of control, with low-level
> input binding the model of environment to reality at one tiny point,
> supplying new facts and showing the mistakes, and low-level output giving
> the model ability to participate in the causal web of environment.

--
Vladimir Nesov
[EMAIL PROTECTED]
http://causalityrelay.wordpress.com/


-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=8660244&id_secret=111637683-c8fa51
Powered by Listbox: http://www.listbox.com

Reply via email to