James Rogers wrote:
> I don't think the smart money is in biological models, though
> it might be where the easy money is.  The model we're using has no
> origins in biology nor do we try and make it have characteristics of
> biological systems -- the idea is to design intelligent systems, not
> clone a brain.  Solving the general problem of intelligence in the
> theoretical sense doesn't really involve knowing anything about
> wetware.  I view it as a case of having an answer that doesn't tell you
> much because nobody knows the question.  For me, checkpointing designs
> against biological systems is mostly a curiosity.

I do have respect for this point of view, but it's definitely more extreme
than my own.

To me, the problem with "solving the general problem of intelligence in a
theoretical sense" is that our contemporary theories of math, CS, and
cognitive science are not up to the task.

Yes, one can make plausible mathematical definitions of general
intelligence, and then prove that given arbitrarily generous computational
resources, general intelligennce is possible.  Marcus Hugger and Juergen
Schmidhuber have done a great job of this in some recent papers, using
algorithmic information theory, statistical decision theory, and related
tools.

But in the (real) case where one doesn't have arbitrarily generous
computational resources, there is no solid well-worked out theory to guide
one... one has to go by intuition, basically.  And it's here that study of
biological intelligence may be inspirational.  If one is trying to build an
intuition about how to create general intelligence using modest
computational resources, doesn't it make sense to look at the best known
example of powerful general intelligence using modest computational
resources?? [i.e. the human brain].  This doesn't mean one tries to simulate
or even closely emulate biological systems, but it does mean that the
comparison with biological systems is more than a curiosity...

Now, James, maybe you really have such a great intuition for AGI design that
you can make all the right decisions without ever looking to biosystems for
inspiration....  Time will tell, as with all our AGI attempts, eh?  ;)

ben

-------
To unsubscribe, change your address, or temporarily deactivate your subscription, 
please go to http://v2.listbox.com/member/

Reply via email to