On 25/09/2014 22:22, Matt Mahoney via AGI wrote:
1. Intelligence depends on knowledge and computing power. A program that rewrites itself cannot gain Kolmogorov complexity. Therefore, self improvement will come from acquiring hardware and learning from the environment.
One problem with this is that short programs can clearly produce high Kolmogorov complexity if given enough runtime. Consider a simple counter. Run it for long enough and it will count from 1 to: 154998464670145002987498798679541316549415641078895144891513. This is a sequence with considerable Kolmogorov complexity. We have compact specifications of intelligence - such as AIXI. Expanding these short descriptions into practical agents is possible, but time consuming - due to the need to search a large search space in order to find them. A deterministic computer program can indeed gain Kolmogorov complexity just by running. The problem is not that this is impossible, but that it happens slowly. Another problem is that we don't care about K-complexity - what we want is agents that can do work for us. Consider a go program, for example. That could win prizes for us and lead to fame and fortune. However, a powerful go program can be specified with low K-complexity. You just write a search program to search through possible go programs until you find one that does very well against possible opponents (weighted by the size of their source code). Here the search program and the halting criterion are both simple - so this is an example of a valuable skilled agent that can be specified with low K-complexity. -- __________ |im |yler http://timtyler.org/ [email protected] Remove lock to reply. ------------------------------------------- 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
