--- On Fri, 12/26/08, Philip Hunt <cabala...@googlemail.com> wrote: > Humans aren't particularly good at compressing data. Does this mean > humans aren't intelligent, or is it a poor definition of intelligence?
Humans are very good at predicting sequences of symbols, e.g. the next word in a text stream. However, humans are not very good at resetting their mental states and deterministically reproducing the exact sequence of learning steps and assignment of probabilities, which is what you need to decompress the data. Fortunately this is not a problem for computers. > If I define intelligence as the ability to catch mice, does > that mean my cat is more intelligent than most humans? That is the problem I am trying to avoid. I now have two benchmarks based on two different definitions of intelligence. One is based on text compression because the ability to model text strings (i.e. compute P(A|Q) = P(QA)/P(Q)) implies knowledge of the distribution of answers A to question Q in written communication, and therefore ability to pass the Turing test. The second test is based on universal intelligence as proposed by Hutter and Legg. The problem with the Turing test is that it only defines human level intelligence. The problem with the universal intelligence test is that nobody accepts it, because nobody thinks anything could ever be smarter than a human. This isn't just an ego problem. It is fundamental. You can't understand a superhuman AI unless you are already that smart. You have no way of knowing whether its answers are correct unless you know them too. Most compression tests are like defining intelligence as the ability to catch mice. They measure the ability of compressors to compress specific files. This tends to lead to hacks that are tuned to the benchmarks. For the generic intelligence test, all you know about the source is that it has a Solomonoff distribution (for a particular machine). I don't know how you could make the test any more generic. -- Matt Mahoney, matmaho...@yahoo.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=123753653-47f84b Powered by Listbox: http://www.listbox.com
