Benjamin Goertzel wrote:
Unfortunately, I do not think any of the existing definitions of
intelligence (include yours above, and those offered by Hutter, Legg,
etc) are worth anything, for the following reason:
Take a look at the word 'goal'. The only way that this term can be
defined is subjectively: you have to use ANOTHER intelligence to
interpret what counts as a goal or not.
Well, one can define an "implicit goal" of a dynamical
system as a function that the system approximately maximizes as
it changes over time...
Then, an intelligent system is one that maximizes complex functions
that are dependent on complex environments...
What's the problem with that?
(In both Hutter/Legg's and my mathematical approaches, btw, we do
formalize the def'n of "goal" in this sort of way. It is not left
subjective
and unspecified.)
What's wrong with it is that this definition is so broad that it can
make a thermostat look like it is an intelligent system.
Any definition that classifies thermostats as 'intelligent' is broken.
Anyone can come up with a definition of intelligence that accidentally
encompasses most of the universe, as well as the real intelligences.
The trick is to find a non-circular definition that leaves out the
thermostats, cuddly toys, Conway's Game of Life and the little program I
once wrote in Fortran that printed out HAPPY BIRTHDAY.
Very old argument.
Richard Loosemore.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
