On Wed, Nov 12, 2014 at 5:15 PM, Mike Archbold <[email protected]> wrote:
> See your point.  But, remember airplanes fly in a 3 dimensional space,
> and an idea/thought always seems to be in the same "place," like a
> 0-dimensional space, which is not a space at all.

Sure, but this just means that intelligence requires a *different*
math than aerodynamics...

Proof-based software validation, and the theory of the complexity of algorithms,
are also based on math --- but not dimensional math...

I have little doubt that in 50 years there will be a pretty developed
math of AGI,
used to refine AGI designs and ideas, much as differential equations
are now used
to refine aerodynamics ideas (every aerodynamics simulator uses the
Navier-Stokes
equations for example) ....

> But, anyway, mathematics is important to AI.  I guess I think there is
> a difference between finding AI *in* mathematics, or using mathematics
> as a processing physical/computational substrate.  I think some
> researchers think in the platonic/math realm it is just a matter of
> uncovering the hidden intelligence INSIDE math.  Need to consider it
> more... not that I have a better alternative at the moment, mind
> you...

Sure...  But whether you're a Platonist or not, the equations of physics still
have the same empirical validity, right? ;) ... the same will be true of the
equations of intelligence once we know them well...

For the time being, we're in the early stages of AGI R&D, so we are proceeding
via a messy mix of math, science, intuition, experimentation and so
forth -- as at the start
of any new science/engineering discipline....    And the mix each
researcher chooses
will depend somewhat on their taste...

-- Ben G


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to