Can Computers Really Think? - Thinking as Computation - page 269

If this is the way one chooses to talk about computers, the discussion stops here.
There is really no point in talking about whether they can think if one is not even
willing to concede that they can add.
This is not a particularly helpful way of talking, however. Imagine we are studying
the bewildering game of Go and wondering whether a computer has ever won against
a Go champion. Suppose we get an answer like this:
No computer has ever beat a Go champion because a computer cannot even play
the game of Go. All it can do is move around meaningless symbols.
This way of talking is not helpful because it does not make the simple distinctions
we are looking for. We want to be able to say that yes, a computer did win against
a chess champion (once), but no, a computer has not yet beat a Go champion. But
in talking this way, we are allowing that computers can indeed do things like add
numbers, solve Sudoku puzzles, and play world-class chess. (You can't win at chess if
you can't even play chess!)
In essence, we are saying that the computer's external behavior (as given by the
inputs and outputs) is enough for us to judge. We do not need to know about the
hidden internal state of the computer or whether there are other interpretations of
the inputs and outputs. Fortunately, this is the way most people choose to talk about
computers.
(As a side note, there is something right about shifting the question about what
computers can do to the question about what people can do. When we ask, “Can a
computer play decent Go?” what we are really asking is, “Is there a suitable computer
program?” that is, a program that when run on a general-purpose computer produces
the external behavior we want in a reasonable amount of time. And to go further, it
is not just a program's existence that matters; we might know that there must be such
a program without anyone's knowing how to write it. Our question is really, “Has
anyone been able to write a suitable program?” So the question is really about people !
When we say that the Deep Blue computer beat a chess champion, and we applaud
that achievement, it is really the team from IBM that produced Deep Blue that we
are applauding. Of course, some day, we may be more willing to give credit to the
machine itself; but that's for another day.)
Now let us return to the original question: Can a computer really think? Can it
understand English? Can it be intelligent? Here we do not seem to be talking about the
computer's external behavior but about its internal workings , not just what a computer
can do, but what it can be.