That's odd, since they have beaten Kasparov (or was it Karpov?).
> Fascinatingly only yesterday I answered another post on a different List
> with the following:
>
> -------------------------------------------------------------------
> I used to play tournament chess, so I have a bit more interest in
> this.
>
> I agree with Searle (I think it is him) who argued that such
> computers are not playing chess at all. They are *simulating*
> playing chess, like a flight simulator simulates flying.
Next, you'll be saying that calculators don't *calculate* at all; they only
*simulate* calculating. This is such an obviously circular position that
even *I* am surprised that you use it. It's true that they are not playing
chess in the same sense that humans do. So what? That's irrelevant. The
*fact* is that they can produce better chess-game results than 99.999
percent of the human population. If you don't want to call it "playing
chess," you are free to adopt your own terms for it, but it doesn't change
the *facts*. If you think it does, then why don't you challenge the Deep
Blue people to play against their machine and see how well you do? The
*fact* is that you will probably be wiped out in every game, especially if
your reasoning in chess is like it is on evolution theory.
Also, will you soon start telling us that clocks don't keep time?
If any or all of these things are only simulations, please tell us in what
*relevant* sense they differ from the real thing? How do you know you are
playing chess when you do, and that the computer *isn't* playing chess? Is
it a matter of being conscious? Fine. Since much of human chess playing is
done subconsciously, I suppose that would mean that *humans* are only
playing chess in some very trivial sense, and only to the extent that they
are *consciously* examining each alternative sequence in detail, rather than
relying on memory and subconscious processes.
> -------------------------------------------------------------------
>
> >CC>More likely, it's evidence that the
> >machines aren't well-programmed for this task. What will you say if, in a
> >few years, machines are able to recognize faces easily?
>
> I would say that they are not recognising faces at all. They are
*simulating*
> recognising faces. Only when they *know* they are recognising faces will I
> admit that computers are *really* recognising faces.
Chris
Ah, but *then* you'll have the perfect *evasion* of reality all ready: They
won't really *know* they are recognizing faces; they will only be
*simulating* knowing that they are recognizing faces.
This is a nice but *very* general escape, since it can be applied to
*anything*, including all human activity.
Well, guess what? When a baby first recognizes a face, *it* doesn't know
it's recognizing faces, *either*. Are you going to claim that babies *don't*
recognize faces, that they are only *simulating* recognizing faces?
If a computer can reliably pick a specific randomly-selected face out of a
random selection of faces, as does a human, on what grounds can you say that
it doesn't recognize faces?
I don't think you know what you are talking about; you are only *simulating*
knowing what you are talking about. For all we know, you only *simulated*
playing chess. How would we know? Just as you only *simulate* knowing
evolutionary theory (and not too well), *everyone* could be only
*simulating* whatever it is they appear to be doing. Writers could be only
*simulating* writing. Computer programmers, artists, music composers, and so
on might only be *simulating* their respective activities, given your
approach.
Stephen
> What will Chris say if, in a few years, machines are *still* not able to
> recognize faces easily?
Chris
Well, since computers *are* used for human identification processes, and
since the software for doing so is getting better and better (sometimes
without further human intervention), I'd say my chances are pretty good that
they *will* be able to do this, even if we have to *copy* the part of the
brain that does this into a computer form.
I will say, also, that you are *still* offering a "God of the ever-dwindling
gaps."
And, I would ask, "*Why* can't it recognize faces well?" I would ask, "What
aspect of human face recognition is missing?"
Finally, I would point out that human face recognition facility is evidence
of *evolution*, not a designer. Facial recognition is important to a species
that communicates largely by facial activity, and upon whose beneficence
*baby* humans depend. Since an infant cannot understand spoken conceptual
language as such, it must make do with whatever information can be
genetically encoded in a generally-useful way. Humans who *can't* recognize
human faces and get a lot of information from doing so tend to get killed
and tend to have relationship problems that will keep them from passing on
their un-social genes.
All designer theory can say about this facility is: "Well, the designer
chose for it to be that way." That's designer theory's necessary answer to
*every* question of this sort, because, lacking any *actual* evidence of a
designer, there is nothing else that *can* be said without going *beyond*
designer theory. In fact, if it were found that humans were especially *bad*
at face recognition, *that* would suggest that evolution was false, just as
finding that tigers had no claws, no sharp teeth, that they had an extremely
strong and distinctive odor, and that they always tried to sneak up on their
prey from upwind while making horrendous screaming noises would tend to
suggest that evolution was false.