🚀 go-pugleaf

I would suggest that we use the word 'sentience' in preference to
intelligence in this discussion.  It more clearly means what we intend to
mean.

In article <7726@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
>In article <1992Oct14.223757.103240@Cookie.secapl.com> frank@Cookie.secapl.com (Frank Adams) writes:
>>In article <7707@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
>>>  (Of course, it may also turn out
>>>that there's no way to get a computer to pass the TT just by
>>>giving it the right program.)
>>
>>I disagree.  My belief that functional organization is all that is required
>>for intelligence/qualia/consciousness is much more firmly grounded; it is
>>essentially a philosophical position.
>
>Fair enough.  But I don't see how anyone can yet rule out the
>possibility that certain low-level physical properties will turn
>out to be necessary.

I have two basic arguments here.  (Note that I will be concentrating on
sentience/thinking here.)  One is that there are any number of things
which we do as part of our thinking (e.g., playing chess), which we've been
able to do with computers simply by creating an appropriate functional
organization.  There doesn't seem to be any distinguishing characteristic of
these tasks from the ones which we have not been able to do, that would
imply that anything more than the right functional organization is required.
Indeed, what these tasks appear to have in common is that they require
a global understanding, not just understanding of a particular problem.

Secondly, I think that sentience and thinking *are* essentially defined
functionally.  Intelligence is the ability to react to one's surroundings
based on reasoning about them.  (Don't tie me to that precise definition.)
Sentience is a certain degree or quality of sentience.  (Exactly what degree
or quality is a research question.)  This is a functional problem, and thus
only the functional organization is relevant to the solution.

>>I expect that the existence of reflective capability will turn out to be an
>>essential part of what constitutes intelligence.
>
>I didn't mean initially to put much emphasis on "linguistic form", but
>I'm not sure there can be any way of representing reasons to oneself
>that we would recognize as such w/o us also seeing it as linguistic
>(where the reasons are, roughly, of the sorts where _we_ use language).

If you extend the concept of language far enough, this is probably true.
But the reflections of an intelligent entity need not be in a language it is
capable of conversing it.  Likewise, I expect that once you extend the
concept this far, there will turn out to be other "languages" used in our
mental processes.

>Anyway, I think we should be wary of thinking reflective = conscious.

Agreed.  Clearly, reflection is not enough to produce consciousness.  I had
only the converse implication in mind.

>>Unlike you, I strongly expect that intelligence, consciousness, and qualia
>>will prove to be irretrievably intertwined, because the type of system
>>needed to produce any of them will necessarily produce all of them.  (That's
>>a nomic "needed" and a logical "necessarily", in case anybody's keeping
>>score :-)).
>
>I think you may be right _for certain meanings of intelligence,
>consciousness, and qualia_, but I think "intelligence" is at least
>headed towards having a meaning that refers only to performance
>(behavior), and not necessarily at a level that would require
>something like consciousness.  Indeed, I think it's already gone
>far enough in that direction so that the answers for intelligence
>and (human-like) consciousness can be different.

I would say rather that intelligence has many distinct but related meanings;
one based purely on performance, one corresponding to my definition above,
and another synonymous with sentience.  However, I think both are still
essentially functional definitions.  Yet another meaning entails
consciousness and qualia.  (Note that by my definition, a gigantic
multiple-choice program would not be very intelligent, since the reasoning it
does is of the most rudimentary sort.)

Note that the moves toward functional definitions of consciousness, like my
definition of intelligence, are not purely behavioral.  Thus they are, I
assert, attempts to explain, not the attempts to explain away that Jeff
quite reasonably complains about.