From: Joaquim Almgren Gāndara (email@example.com)
Date: Sun Dec 03 2000 - 11:37:27 MST
> Sophisticated methods of generalization do not necessarily abstract more
> data. Rather, sophisticated generalizations abstract irrelevant data out
> while leaving relevant data in. If you abstract relevant data or leave in
> irrelevant data, then the method is unsophisticated.
I think I see your point. But the fact remains that we have some kind of
trade-off -- we either have errors due to generalisation or rigid stupidity due
to overfitting. I'm afraid that the upper limit to intelligence is the optimal
ratio between these two. (This is where you come in and tell me what the third
aspect of intelligence is that solves the problem and can raise the level of
intelligence regardless of how good/bad generalisations it has.)
> You can't induce a limit on intelligence
> from your observed experiences with data fitting - you are, as it were,
> abstracting relevant context and leaving in irrelevant data.
Hmm. I hope you're right. In fact, that was a very good argument. However, my
proposition was that if generalisation is a fundamental building block in
intelligent systems and the only way to become more intelligent is to make
generalisations with lower error rates, then the intelligent system will
theoretically reach a point where the error is zero or too low to allow for
This proposed upper limit to intelligence poses these two questions:
* Is the error caused by abstracting more or less irrelevant data the only
variable that affects the level of intelligence?
* Is above mentioned error necessary for creative thinking?
When I'm convinced that the answer to those two questions are "no", I'll be
convinced that generalisation does not cause an upper limit to intelligence.
- Joaquim Gāndara
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:35 MDT