On Sun, Nov 23, 2014 at 8:25 PM, Alan Grimes <[email protected]> wrote:
> An AGI is a **LEARNING SYSTEM**. Learning is it's sine qua non. If it
> can't learn, then it isn't an AGI, once you are at that point, all you
> need to do is connect a few network cables and walk away.

I don't understand why you haven't created an AGI yet, if it's so simple.

Oh wait, it's not. Are you familiar with Legg's proof that there is no
such thing as a simple, general purpose learning algorithm? Powerful
learners are necessarily complex. http://arxiv.org/abs/cs/0606070

In case you don't want to read the paper, here is a simple proof.
Suppose you have a small program with Kolmogorov complexity n that
learns to predict bits in arbitrary sequences. Then I can create a
sequence with Kolmogorov complexity slightly greater than n that your
program can't predict. My program simulates your program and outputs
the opposite of whatever your program predicts.

-- 
-- Matt Mahoney, [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