Hi. Thanks for your interest, Danko. My two cents is:

"No Free Lunch" theorems are weakly associated with Legg's proof -
since both are to do with limits on optimization theory. However,
"No Free Lunch" theorems are not much of a barrier to machine
intelligence - since we know that the world is structured via
Occam's razor, and so optimization problems we might face are
not drawn form a uniform distribution.

Regarding the idea that human understanding represents a barrier
to development of machine intelligence - again, there's obviously
some truth in this (or we would be done by now) - but the
understanding of one human is fairly easily overcome via
collective intelligence by slapping a bunch of them together -
or by augmenting their intelligence using machines. So there
isn't really much of an absolute barrier there.

Machine intelligence headlines are dominated by neural
nets today - but it is quicker to scale via making multiple
copies than it is by trying to make a neural net bigger.
That means that we will have a society of agents - and
that their rules of interaction will be important. We see
this today since we have computers and we have the internet -
and the internet is quite important to the overall dynamics.

The rules of evolving societies are what students of
cultural evolution look at - so hopefully this fledgeling
science will be able to contribute to the political and
economic processes that will shape civilization long
after we have squeezed what juice we can out of making
bigger and better neural networks.

--
__________
 |im Tyler http://timtyler.org/ [email protected]



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to