What follows is not part of this paper per say. But, I thought it would be interesting to get a collection of attitudes or ideas about machine intelligence from professors, students, and people not associated with computer science. I do not think any of the later answered me.
All professor and student references, unless otherwise noted, are from the University of New Mexico. Below is the e-mail I send out. On the following pages are the replies
Hello, I am writing a paper on Alan Turing and would like to end it
up with a sort of informal pole of attitudes or ideas.
If you have time could you answer the question?
Can machines think?
Thank you
Strider
********************************************
Reply from xxxxxx xxxxxxx - PHD Computer Science
Sandia Labs
***
Yes.
In my opinion, computer chess is a prime example of this. Chess used to be
a benchmark for whether computers could display intelligence. In the
beginning people would say that if a computer could play chess at the level
of a beginner then it would have displayed intelligence. Then when computers
played that well, they raised the "bar" on what intelligence meant -- they
said something like "if a computer could only play C level tournament chess,
then it would display intelligence", then "if a computer could only play a
XX level, then it would display intelligence", and so on and so on. Well,
they kept raising the bar and finally it boiled down to "if a computer could
only beat every last person on the planet in chess, THEN it would display
intelligence" --- guess what?
Imagine if you could take deep blue (the chess machine that beat Gary
Kasparov--who by many is considered to be the greatest chess player of all
time) back in time to, say, the 50's. What do you think those people would
have thought about this machines ability to think? They would have been
absolutely astounded at this machines ability to play chess. Descriptions
like "beyond belief", "absolutely astounding", "unquestionably brilliant"
would have been thrown around.
My feeling is that what qualifies as "intelligence" is a sliding scale. At
any given point in time, machine intelligence, almost by definition, is just
beyond the horizon (where ever that horizon happens to be). A similar thing
holds for science and magic. Things you don't understand historically have
been called magic, and things you do understand are science. So a similar
question can be asked "does magic exist?" After all, intelligence is
something we don't understand. Think about it.
xxxxxx
************************************************
Reply from xxx xxxxx - Computer Science undergrad.
***
can machines think?
do chickens have lips?.......
Seriously, I don't see how. Machines have no soul, no morality and no
innate since of right and wrong. God made humans. Humans made machines.
Remember our conversation............A machine can write the perfect song.
But when it plays it, it doesn't know whether it's good or not........
xxx
circa 1999
**********************************************
Reply from xxxxxx xxxxxxx - Computer Science undergrad.
***
Heck No! Most of the people I see driving on the road in the morning don't
seem to be thinking. Why would a computer be able to? Besides, all the
computers I've met pretty much do exactly as the program indicates, erroneous
or not. Without free-will, there is not thinking. Hence Republicans and
Democrats do not think and neither do their computers.
xxxxxx
********************************************
Reply from xxxxxx xxxxxxx
- Computer Science professor at San Diego State University
***
I'm afraid that is a research paper not an email message.
It does not lend itself to a Yes/No answer.
xxxxxx xxxxxxx
**********************************************************
reply from: xxxxxx xxxxx - PHD candidate
***
What do we mean by think? If we ask the question, as Alan did, from a
behavioral standpoint, then we can assert (as the yearly Turing Contest
attempts to) that in certain rich, nontrivial, yet restricted domains
machines have been demonstrated to be indistinguishable from people. If
we use the recent decades of research to refine the question and ask, "can
machines create and discover in themselves capacities for dealing with
unfamiliar situations and build on those creations and discoveries?", then
we may be forced to answer: perhaps just barely and almost certainly
eventually.
Whether through design by their creator or through rapidly engaged
self-evolution, we are beginning to construct artificial
systems--implemented in machines--rich enough to qualify as complex
systems. Nature and the study of complex systems has shown us undeniably
that above a certain "critical mass" of complexity systems begin to
evolve. This is an auspicious moment in history because this has begun to
be possible. What remains to be seen is whether we, as the progenitors of
these alien new systems, have imbued them with the appropriate initial
tendencies that will allow them to grow effectively and become...thinkers.
************************************************************
reply from: xxxxxx xxxxxxx - Professor Computer Science
****
Unfortunately my ability to answer that question does not really
depend on whether I have time or not.
I can only answer it with another question: Do you think a person is
some kind of machine? If not, why not.
--
xxxxxx xxxxxxx UNM Computer Science
***********************************************************
reply from: xxxxxx xxxxx - Professor Computer Science
***
It is perfectly possible for machines to "think". One problem with this is
what we mean by "think". What Turing did with his Turing test is find (or
attempt to find) and objective, scientific way of saying a machine (or alien
being or anything) is "thinking". I certainly believe that someday a machine
will pass Turing's test. No machine is even close now, however. I don't have
any opinion of whether a machine can have a "soul" or whether a machine
can "think" if we attempt to define think in human-centric and non-objective
terms.
**********************************************************
reply from: xxxxxx xxxxxxxx - undergrad Computer Science
***
Hey dude. Sure, he's my take.
No, machines cannot think. If we were to look at the
definition of "think" (and I don't have dictionary in
front of me) I would say that the word "think" would
say something about reasoning. Maybe another way to
reword the question, can machines reason? Now, what goes
on in your head when you reason. Morality and ethics come
into the picture I believe. Machines can have morality and
ethics programmed in then. What about emotions? Do emotions
have an impact on reasoning? Yes they do. Can a machine
have emotions? I would say it is highly unlikely. Can you
reason effectively without emotions? I would think that
if a person had no emotions what-so-ever then that person
would not have morality or ethical issues. Is this safe?
My take on this is that machines can compute or calculate
which may (on the surface) appear as though they are thinking.
Machines can make decisions based on programming and / or
previously seen calculated results. Imainge this, do you
believe that you could ask a machine what it's favorite color
is? Hmm, says the machine, let me think. Infinite loop!
But who am I to say this, I'm sure there are many people
smarter than me that know more about this subject. I am
still a "pee-on" in the computer science world.
**********************************************************
reply from: xxxx xxxxxxxx - undergrad Computer Science
Madcat group member
***
I think they can think, but not in the traditional AI way, and I think
they're still pretty dumb.
I feel that chess-playing programs and other AI programs that basically
search a tree do not think the same way we do because I think we very
rarely spend our thinking energy trying to figure out a series of
formalized steps to solve a problem.
We can do amazing feats of pattern processing and world model building.
Think about how think about and visualize your route home every day. You
can get there, you intimately know the route, and yet you don't have to
model every piece of gravel or sand on the shoulder of the freeway. How
do you do it? How do you decide what is important in your world? How did
you learn anything - no one taught you or showed you the search space.
What kinds of things do you store in your brain? How do you feel about
your immune system's ability to pattern-match and think inside your body
completely independently of your brain?
I think some of the slippery architectures like neural networks have
potential to think about and process the world in a way similar to the way
we do, but right now, I think they are still less-cognitively-cool than
dust mites.
I think machines are capable of thinking, but they aren't too smart yet.
**********************************************************
reply from: xx xxxxxx xxxxx - undergrad Computer Science
***
Well I think its possible. I think the traditional architecture of
computers may limit that ability right now. In order to think, we need to
use the only model we have and that is the brain. The brain has no silicon
or copper, and all though we may know what parts control which organs and so
forth, we have yet to really tap what an emotion is. Is it all the cells in
one region getting stimulated, or is it a chemical produced? I think that
thinking and emotion go hand and hand, they are not the same, but to think
about something you generally have to have an opinion or two about it. So
yes, I think its possible, but we need to expand the traditional VN
architecture model.
**********************************************************
reply from: xxxxxxxxxx xxxxxx - professor Computer Science
***
It depends what you mean by think.
Can machines think now? Probably not.
Will they be able to sometime in the future? Probably yes.
**********************************************************
reply from: xxxxxx xxxx - graduate Student
***
maybe.
**********************************************************
reply from: xxxxxxx xxxxx - PHD engineering
Sandia Labs
***
My short answer is that with current technology, machines cannot think. But I
would accept the possibility that (still undeveloped) technologies could
permit mechines to think. This is a VERY philosophical question.
**********************************************************
reply from: unknown - (forward from xxxxx xxxxxxx)
***
Here is a response to your question from one of my computer science
friends
Since I do most of my research in AI, I suppose I should have a
stock answer for this question. Not to sound too much like Bill
Clinton, but it depends on what you mean by the word "think".
Computers will eventually be able to emulate most human responses
to situations. Currently, they lag behind us badly in certain areas,
the most notable being common sense and abstraction. In other areas,
they excel beyond our capabilities (e.g. raw computation).