On Mon, 10 Sep 2001, Mike Lorrey wrote:
> Party of Citizens wrote:
> >
> > On Mon, 10 Sep 2001, J. R. Molloy wrote:
> >
> > > Define intelligence as the ability to solve problems and answer questions
> >
> > How much of Hawking's work entails solving problems and answering
> > questions in mathematics? How much of that can be programmed into machines
> > now? Why shouldn't Cambridge replace Hawking with a robot? See why hemight
> > be worried?
>
> Moreover, intelligence, contrary to JR's assumption, is not the be-all
> and end-all of sentience, though it is a part. Creativity, for example,
> and the ability to make the value judgements that filter 'good'
> creativity from 'bad' creativity, are a hallmark of human sentience.
> When a computer can look at another sentients work and say "That
> is/isn't art", with some rational explaination for it, as well as create
> art that a) is not derivative, and b) critics can say IS definitely art,
> then I'll accept that AI computer as sentient. It cannot do so without
> values.
I think you hit upon several of JR's useless concepts above. Scientific
psychology began exorcising them after 1879 and by the time of Watson's
seminal 1913 Psych Review paper, the ghost in the machine had been
thoroughly exorcised. If it cannot be reliably observed and therefore
measured it aint in the domain of science. AI researchers often chase
after ghosts because they haven't studied real human intelligence and
especially the work of 5 generations of smart people who dedicated
lifelong careers to spelling out observable-measurable-testable
intelligence. If you discover some kind of observable-measurable-testable
intelligence which is significant and hasn't been reported since 1879,
then rush to get your ideas published in a psych journal. You will be
famous. It is as likely that this will happen as the discovery of a new
species of mammal at this time. But don't give up...Sasquatch hunting is a
big tourism draw in BC.
Meanwhile, if you have a machine which gives the answers/solutions a la
JR's definition (which is as good as any) then it meets the human
equivalency criterion. Period. Now let's go back to Hawking. All kinds of
math-like tests are used to assess human intelligence. Is there any reason
a robo-mathematician couldn't answer all question and give all solutions
as well as Hawking?
POC
This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:40:28 MDT