Volume 3, Issue 3
3rd Quarter, 2008


The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind

Marvin Lee Minsky, Ph.D.

This article was adapted from a review Marvin Lee Minsky, Ph.D. presented on his book, “The Emotion Machine”, during his attendance of the 3rd Annual Colloquium on the Law of Transbeman Persons, December 10, 2007, at the Florida Space Coast Office of Terasem Movement, Inc.

Dr. Minsky explains the many different and resourceful ways and circumstances in which the human mind thinks and how we may build machines to think as humans do.

I am going to talk about why AI, artificial intelligence research, at least from the outside, seems to have gotten stuck and what we can do about it.



This is a list of problems that people are going to have to solve soon, and before meeting Martine Rothblatt [1], I hadn't actually thought about the problem of merging or resolving the possible conflicts between artificial beings and the ones we have now. I figured that when the machine started to get smart people would start to think about it, but it might be really nice to start thinking about it when we have a few years to prepare. Therefore, this is just a list of typical problems and solutions to them. We are going to have to solve all of these problems little by little, but one of the most serious problems is the one that will be caused by the drive toward extension of life and ultimately immortality because this is going to cause a serious labor shortage as soon as it starts happening.

As it is commonly known, the expected length of human life at present, in the developed countries, has been growing at the rate of about one year every four years or three months per year. People being born today are living twelve or thirteen years longer than people born in 1960. Now the strange thing is that the maximum life span has not changed for a very long time. Every now and then somebody lives to the age of 120, and there is no validated incidence of anybody living to 125. Jeanne Calment reached 122.

I read in an interview, she met Van Gogh and they asked her, "Well, what did you think of Van Gogh?" She said, "Unfortunately, he had such bad breath that I couldn't stay in his presence long enough." It was very valuable to have this information because the biographers were too polite to have written it down.

What is going to happen is this: There will be one young person for every fairly old person. I am assuming that as the lifespan approaches 180 people will remain in excellent health. I will bet most of you have looked at the predictions by Aubrey de Grey.[2]

Aubrey De Grey has this rather complicated theory about the improvements in health being able to catch up with the progress of decline of the fragility of people. We will see if that is true but he thinks that many of the people being born today will be in the era where diseases and degenerations are being conquered at a higher rate than they occur so that there is a moderate chance of living forever. That's all very nice, and one problem is: who will do the work? I think this problem is serious enough that in just another twenty years or so we will be suffering a labor shortage.

Many developed countries are already suffering an acute labor shortage: most of Europe, United States, Japan, so forth. Instead of robots, they are getting migratory workers, guest workers, and so forth. The problem of citizenship is becoming acute. I am very sympathetic to the Western cultures, and the Eastern culture and so forth, who are terrified that the immigrants will destroy the culture they have. I don't know that all cultures are equally good but I sympathize with the problem. People do not talk about that, but cultures take a long time to develop and some are, in my view, much better than others are. The greatest culture of all is the one called science and how do we know that science will survive? It has disappeared in the past. Archimedes was on the verge of doing what Newton did and for some reason mathematics and science stopped for a thousand years. It might be about to happen now, in which case forget these curves.

Why don't we have artificial intelligence yet? There were some remarkable achievements in very early days of artificial intelligence. This is my favorite one, a young student who happened to be blind named James Slagle [3] wrote a Ph.D. thesis in 1961 that was nearly as good as a good MIT freshman at doing intergroup calculus. Up until then there was no general theory of how to integrate functions. Isaac Newton invented the process but could not solve it. People like Gauss [4] and others spent the next couple of centuries on it. By 1950, there was a great collection called the Bateman Manuscript Project [5] run by the American Mathematical Society and they collected integrals.

What's the integral of that? They had reached something like several hundred thousand integrals kept in this building, mathematicians would send them in, and the Bateman Manuscript Project was getting big. Now, Slagle did not solve that problem but he was followed by another mathematician graduate student named Joel Moses [6] who made a somewhat better program. Incidentally, it took this IBM701 about fifteen minutes to solve this problem, which is how long the MIT student took to solve that kind of problem if they were successful. This number has absolutely no significance; it is just a remarkable coincidence because, as you know, computers are ten million. The 701 ran at 20,000 instructions per second, and for a couple thousand dollars, you can buy a floor processor that runs at 20 billion. It was a funny number and it made us feel the damn thing was life-like because it was so slow. By 1969, these processes got the attention of other mathematicians and the general problem was solved. By 1969 Bobby F. Caviness and someone else finally wrote down the procedure which can integrate any expression that has an integral.

It is 1961 and this thing is doing as well as college students. We were very impressed with that but, of course, it could only handle the formal problems on the exam. Then three years later Danny Bobrow,[7] another graduate student, ex-chemist, wrote a program that can do problems, word problems, in high school algebra. Fewer college students can do those in calculus for some reason, sort of tricky.

Bill's father's uncle is twice as old as Bill's father. Two years from now Bill's father will be three times as old as Bill. The sum of their ages is 92. Find Bill's age.

Bobrow’s program knew barely enough about the meanings of a few English words to figure out which of those were mathematical expressions. It was assumed that every sentence was an equation. If you have three sentences, it would have simultaneous equations and it got pretty good at that. By 1970, we had robots that could solve geometric problems.

Next Page

Footnotes

1. Martine Rothblatt, J.D. Ph.D. - started the satellite vehicle tracking and satellite radio industries and is the Chairman of United Therapeutics, a biotechnology company headquartered in Silver Springs, Maryland. Dr. Rothblatt is also the President of Terasem Movement, Inc. and has written several books, including The Apartheid of Sex, Two Stars for Peace, Unzipped Genes, and Your Life or Mine.
http://terasemcentral.org/about.html July 14, 2008 4:44PM EST

2. Aubrey de Grey, Ph.D. - Chairman and Chief Science Officer of The Methuselah Foundation. His major research interests are the role and etiology of all forms of cellular and molecular damage in mammalian aging, and the design of interventions to reverse the age-related accumulation of such damage. He has published extensively on these and other areas of gerontology, and is also Editor-in-Chief of the high-impact journal Rejuvenation Research, the only peer-reviewed academic journal focusing on intervention in aging.
http://www.methuselahfoundation.org... July 14, 2008 4:55PM EST

3. James Robert Slagle – a Freshman Calculus Student and Ph.D. candidate at the Massachusetts Institute of Technology who, in 1961, wrote a dissertation entitled:
Heuristic Program that Solves Symbolic Integration Problems in Freshman Calculus, Symbolic Automatic Integrator. To see this dissertation, go to:
http://portal.acm.org/citation.cfm?id=321193 July 15, 2008 1:15PM EST

4. Johann Carl Friedrich Gauss (1777-1855) - made the first table of integrals, and with many others continued to apply integrals in the mathematical and physical sciences. Cauchy (1789-1857) took integrals to the complex domain. Riemann (1826-1866) and Lebesgue (1875-1941) put definite integration on a firm logical foundation.
http://integrals.wolfram.com/about/history/ July 15, 2008 1:29PM EST

5. Bateman Manuscript Project - a major effort at collation and encyclopedic compilation of the mathematical theory of special functions. It resulted in the eventual publication of five important reference volumes, under the editorship of Arthur Erdélyi.
http://en.wikipedia.org/wiki/... July 15, 2008 1:32PM EST

6. Joel Moses - Dr. Moses holds a Ph.D., which he received from MIT in 1967. He has served as MIT’s Provost, Dean of Engineering, Head of the Department of Electrical Engineering and Computer Science (EECS), Associate Head of EECS, and Associate Director of the Laboratory for Computer Science. Dr. Moses served as ESD's Acting Director from December, 2005 through November, 2007.
http://esd.mit.edu/Faculty_Pages/moses/moses.htm July 15, 2008 1:43PM EST

7. Danny G. Bobrow - a Research Fellow in the Intelligent Systems Laboratory of the Palo Alto Research Center, and a member of the Natural Language Theory and Technology Area. He received his PhD in Artificial Intelligence from MIT, and has over 100 published papers and books.
http://www2.parc.com/isl/members/bobrow/ July 15, 2008 2:34PM EST

1 2 3 4 next page>