From: Indriunas, Mindaugas (inyuki@gmail.com)
Date: Thu Jun 08 2006 - 07:02:31 MDT
Intelligence
When I was a child, and was thinking of Artificial Intelligence, as a
thinking entity. I always ended up, that it must be just a very simple
principle, an algorythm that, if applied on information, becomes an
aware entity.
Recently I remembered my first dream in my life - to understand
everything. I understood before I had this dream, that the main
feeling I had was endless curiosity.
Then I found that some people in the world also identified themslves
compared to others, as nothing more than curious people.
At this point, I had thought of a hypothesis, that the ultimate goal
of any intelligence, as a "thinking entity", is purely - to understand
everything.
Definition might be obscure, as the word "to understand" is obscure,
and even proposed as indefineable in Wikipedia, but I think it is
sufficient to create a computer program, which would do exactly that,
not necessarily being aware of that at the beginning.
Presently, the definition of intelligence, that I am thinking of, is as follows:
-----------------------------------------------------------------
Passive intelligence is any entity that:
(three conditions for any intelligence)
1. routinely performs comparisons of the streaming data from world
(environment) with the data from the memory,
2. routinely guesses (predicts) possible correlations in data by
empirically estimating functions, by
* routinely intercomparing the data generated with these hypothetical
(guessed) functions and the streaming data from the world
3. identifies functions, which produce the data, as closer to the existing data.
Active intelligence has another feature. It does:
(fourth condition for active intelligence)
4. Try to change the world to get new data.
-----------------------------------------------------------------
Here I think all the parts are quite easy to realize except one - the
guessing mechanism...
but we have many mathematical extrapolation, and prediction methods in
contemporary mathematics.
Generally, it should be one very simple and universal prediction
algorythm, which would recursively reuse the predicted functions to
make new guesses.
Inyuki
inyuki@gmail.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:56 MDT