Structure of AI (was: Re: [extropy-chat] COMP: Distributed Computing)

[J. Andrew Rogers]

On Nov 20, 2004, at 1:02 PM, Damien Broderick wrote:
> I'm not certain, but I suspect that Humberto Maturana and Francisco 
> Varela, among others, were there many years ahead of you (although in 
> cogsci and biology rather than AI).


/me googles furiously

 From what I can gather, their models have some vague similarities but 
makes no assertions of the type I was referring to.  The only thing 
that led me to believe that the notion that intelligence is not 
mathematically bound in any significant sense by computational 
complexity was new, ignoring Penrosian theories of a "something else", 
was the uniform reaction to the idea across a broad swath of AI people 
who I would expect to be familiar with the idea in the abstract if 
someone else had promoted the idea.  It has simply been assumed that 
computational complexity is the bounding problem on normal machinery.

There are two aspects that together distinguish the idea: that 
intelligence scalability is not bound by computational complexity AND 
that this is true on ordinary finite state machinery without invoking 
special physics, spirituality, or anything else.  In the past, the only 
people who generally asserted that intelligence was not a function of 
computation also asserted that this was because there was ineffable 
"magic" of some sort that was generating the intelligence.  In other 
words, they simply delegated intelligence to something else.  My 
assertion was that computational complexity is not a limit to the 
expression of intelligence in a *conventional* finite state machine 
context.

Or to put it another way, there is no theoretical reason that one could 
not create a human-level intelligence on an old 486 computer.  The 
speed of the computer is orthogonal to the problem space, and only 
weakly relevant as a practical matter.


j. andrew rogers