Could you tell us a little bit why are you interested in AI as a sociologist?
- There were several reasons why I am interested in AI.
- I wanted to advance the field of AI.
- I am lazy, and have been since a child. I want 100% unemployment, but with full production of goods made by robots. Perhaps my contest will speed up the process. (The difficult question is how to equitably distribute the goods that the robots produce.)
- I am interested in chaos theory - a small change in initial conditions makes a big difference. For example, the weather is chaotic. A butterfly flapping its wings over Brazil might effect the weather over Paris a week later. It seemed to me that society is chaotic. Before I started the Loebner Prize, there was no Turing Test contest. Anybody could have started the contest and it would have his or her name. Now, the contest is known world wide as “The Loebner Prize.” And I've always wanted to be a social butterfly.
- At the time that I started the Loebner Prize my company, Crown Industries, Inc (http://gocrown.com) was making roll-up plastic disco dance floors. I thought that it would be interesting to have funding for AI come from the sale of plastic disco dance floors, and might popularize them. I no longer manufacture portable dance floors, but the contest continues.
Could you explain why ALICE - the most effectiveprogram
that engages in a conversation with a human - is unable to pass the
- What do you mean by “effective”? In any case, at the last competition, ALICE came in last among the Final Four, so it's presently not the most “effective” program. Rollo Carpenter's program “Jabberwacky” won the contest. However, none of the programs came close to passing the test. I think that is because they are using “canned” or preprogrammed responses to questions, rather than trying to “analyze” the questions of the judges and construct answers that are relevant.
How is the Turing machine is built?
- A “Turing Machine” and “The Turing Test” are completely different things. A “Turing Machine” is a theoretical construct which Alan Turing invented to analyze what a computer is. It's not a real machine, its a theoretical idea of a machine which makes theoretical marks on a theoretical paper tape according to a set of rules. To be honest, I really don't understand the idea very well.
Why the Turing test is
to be an authoritative exam of machine intelligence?
- The Turing Test is authoritative in the sense that it is completely objective. If a human can think, and a judge can't tell the difference between a computer and a the human, then the computer must be thinking also. It's an “operational definition.” (The human is thinking, the computer is doing the same thing, therefore the computer is thinking) However, it is by no means universally accepted as the best test of machine intelligence.
Do you regard that it is possible to create ”thinking machines”?
- Yes. I believe that Claude Shannon, the inventor of information theory was asked the same question and he responded something like:“Yes. I'm a machine and I think.” I agree completely. I think, but I am not alive. I am nothing more than af complex chemical reaction. “Life” doesn't exist. Science requires parsimony – the elimination of unnecessary concepts. If you read Science or Nature or any science journal, they never invoke “Life” to explain something (even though they annoyingly use the term all the time). The fact that I am not alive doesn't mean that I don't have a gigantic ego, or that I don't lust after women, or don't enjoy sushi. I may be only a chemical reaction, but I'm a wonderful chemical reaction, one you'd be proud to bring home to introduce to your mother.
How do you see the NPL development perspectives?
- I don't know very much about the current state of NLP. (Natural Language Processing) in the academic world. The entrants in the Loebner Prize are using programs with a large repertoire of standard responses, but the do not seem to have much (if any) ability to “analyze” the judge's questions and respond based upon the information in the question rather than just as a “canned” response to a recognized statement.
Why do people want to believe that a machine can think?
- I don't think this is necessarily true. Many people refuse to believe that a machine can think. On the other hand, if a computer can simulate a human, and humans are trained to think that others are thinking, then the machine is doing what the human is, so it must be "thinking." This is the essence of Turing's test.
In any case, am unable to understand how or why most other people think what they do. They are, in general, so wrong about so many things. How could anybody have voted for George Bush? I simply can't understand that.
The malicious say that 40% of the human population is not able to pass
the Turing test. Why then should we check that the machine can
pretend a person with define mental predispositions?
- I have never heard this. It implies that against that 40% of the population, a computer would be judged the more human. I don't think this is correct. However, it does suggest that one should choose “intelligent” humans as the criteria against which the computers must be judged.
What do you think about the opinions proclaimed by Marvin Minsky who said that “
I do hope that someone will volunteer to violate this proscription so that Mr. Loebner will indeed revoke his stupid prize,
save himself some money, and spare us the horror of this obnoxious and unproductive annual publicity campaign.” ?
- His remarks were cruel and unnecessary. However, he did spell my name correctly. Actually, I was pleased to be able to respond that his offer made him a sponsor of the contest. I demonstrated logically that his offer of $100 to anyone who could get me to end the contest meant that he was a co-sponsor of the event since the contest would be discontinued when someone won the Grand Prize and he was therefore offering to reward the (future) Grand Prize winner. It's not often that one can one-up Minsky.
- I am more concerned by the fact that Dr. Stuart Shieber, ( http://www.eecs.harvard.edu/~shieber/ ) who was a participant at the first Loebner Prize contest, and who wrote an article about that contest, ( http://www.eecs.harvard.edu/shieber/Biblio/Papers/loebner-rev-html/loebner-rev-html.html ) does not see fit to mention the Loebner Prize in his recent book on the Turing Test despite his participation and his article. I must ask myself whether I have expended 15 years of effort and perhaps USD 200,000 on a worthless effort. Perhaps you might want to contact Dr. Shieber and solicit his opinion. I think his response would make your article more interesting.
- John Sundman, in his article about me in salon.com ( http://archive.salon.com/tech/feature/2003/02/26/loebner_part_one/ ) asserts that the A.I. Community “loathes” me. Perhaps Dr. Shieber's refusal to mention the contest is a manifestation of such loathing.
What do you see in the future for the Loebner Prize Contest? Any predictions?
- Unless some organization takes over the competition it will probably end with my death. I won't be able to run it from the grave. I started the contest in 1990 at the suggestion of Dr. Robert Epstein who was then Director of the Cambridge [Massachusetts] Center for Behavioral Studies (CCBS),. I had previously told him about my idea for a Turing Test Competition and he urged me to let his organization handle the matter, which I did. At that time, I insisted that the contest be held annually. The CCBS agreed.
- Dr. Epstein left the CCBS in 1993 His successors were no interested in the competition. In 2000 The Board of Directors ordered the Executive Director to have nothing to do with the contest. Eventually, I sued the CCBS in Federal Court over their refusal to hold an annual completion as we had agreed. We settled out of court, but the settlement amount was not sufficient to underwrite the creation of a non-profit to organize to conduct the competition. If you are interested the court papers are available for inspection. They are rather interesting (well, at least I think so).
What is your vision of the future?
- "Man is the only animal that laughs and weeps, for he is the only animal that is struck with the difference between what things are and what they ought to be." William Hazlitt
- I don't often think of the future, but when I do my view is grim. There is no preparation for a world of 100% unemployment because of automation. Automobiles are wrecking the planet with their fumes, and their insatiable thirst for gasoline. The greenhouse effect will melt the polar ice caps, flooding coastal areas of the world. And, of course, I'll be dead.
What, in your opinion, is going to be the role of AI within the next years?
- I don't know what it will be. I assume that fewer and fewer people will be employed as automation replaces humans with machines.