Featherless Bipeds

Dinner with Simon

Featherless Bipeds

This featured "Dinner with…" series builds on the classic thought experiment: "Which 5 historical figures would you invite to dinner, and how would you seat them?" While the field of astrobiology historically rests on many "shoulders of giants" –too many for one dinner party, the Astrobiology Magazine has selected some initial candidates for our dinner party, and then asks them to introduce their area of expertise in a brief question and answer format.

The answers are their own, as gleaned from some of their most famous, controversial, or seminal contributions to science and technology. In many cases, the selection of commentary is driven by the curiousity to understand these great historical figures as one might imagine them as more modern characters, perhaps joining in on table talk or an informal interview.

Tonight’s dinner introduces Nobel Laureate, Herbert Simon, widely considered the father of artificial intelligence. As Ronald Marks, a senior analyst with the SAIC Strategies Group, wrote about Simon: "Speaking as the economic ‘everyman’, I believe our new Internet Age will continue to make Herb Simon look like the genius he was."

simon_artificial
Herbert Simon, widely regarded as the father of artificial intelligence research

Today also commemorates Simon’s birthday, June 15 [1916]. Simon was Richard King Mellon University Professor of Computer Science and Psychology, Carnegie Mellon University, and 1978 Nobel Laureate in Economics.

In 1975, he earned the prestigious A.M. Turing Award for his work in computer science. He also was inducted into the Automation Hall of Fame because of his pioneering work in the field of artificial intelligence. Simon was educated in political science at the University of Chicago (B.A., 1936, Ph.D., 1943). He held research and faculty positions at the University of California–Berkeley, Illinois Institute of Technology, and Carnegie Mellon University.

During Christmas break in 1955, he, Allen Newell and programmer J.C. Shaw made their vision a reality by creating Logic Theorist, a computer program that could discover proofs of geometric theorems. It was the first computer program capable of thinking and marked the beginning of what would become known as artificial intelligence. The story goes that one day Herbert Simon announced to a group of his students that he and some of his colleagues had invented a "thinking machine". He said it was equal, and perhaps superior, to the human brain. At that time the computer was chiefly famous for having been used by the British to decode enemy messages in the second world war.

He and Newell, whom he had met at the Rand Corporation in 1952, also wrote the world’s first chess program, although it did not play too well. In 1957, Dr. Simon became convinced that computers not only could think, but that a computer would be able to beat the world’s best chess player within 10 years. It was a prediction that would later come back to haunt him; it actually was 40 years before IBM’s Deep Blue would win the chess championship from Garry Kasparov.


Astrobiology Magazine [AM]: Many astrobiologists wonder about how a future space probe might discover something that is ‘alive’. In what ways has your work in artificial intelligence revealed a computer program–say, a virus–as alive?

Herbert Simon [HS]: Symbol systems solve problems by generating potential solutions and testing them, that is, by searching.

The two most significant classes of symbol systems with which we are acquainted are human beings and computers.

The system also contains a collection of processes that operate on expressions to produce other expressions: processes of creation, modification, reproduction and destruction.

AM: Those four processes are often quoted as common to a thing that meets the broadest definition of ‘life': created, modified, reproduced and destroyed. But is there more to what is biological life?

HS: We must be careful about equating ‘biological’ with ‘natural’. A forest may be a phenomenon of nature; a farm certainly is not.

To say that an astronaut, or even an airplane pilot, is obeying the laws of gravity; hence is a perfectly natural phenomenon, is true, but its truth calls for some sophistication in what we mean by ‘obeying’ a natural law.

Unfortunately the term ‘artificial’ has a pejorative air about it that we must dispel before we can proceed. Our language seems to reflect man’s deep distrust of his own products.

My dictionary defines ‘artificial’ as, ‘Produced by art rather than by nature’.

AM: So very smart computers–say, like Sir Arthur C. Clarke’s HAL in the movie 2001: A Space odyssey, those would be classed as not only alive, but also intelligent life?

Extreme Explorers' Hall of Fame
The autonomous Antarctic meteor finder, Nomad, uses artificial intelligence to recognize and classify promising rocks
Credit: Carnegie Mellon, cmu.edu

HS: Any physical symbol system of sufficient size can be organized further to exhibit general intelligence. By general intelligent action we wish to indicate the same scope of intelligence as we see in human action.

Computer programs are capable of making actual discoveries that model important cases from the history of science.

AM: Clarke is credited with describing humans as ‘carbon-based bipeds’. At least, that defines homo sapiens as a species might be observed by zoologists from afar. You coined a different description.

HS: All humans are featherless bipeds.

AM: But doesn’t there have to be a piece of that definition that includes ‘intelligence’. What kinds of decision-making seem unique to us terrestrial bipeds? And what’s not artificial?

HS: [In the 50's] I felt increasingly the need for a more adequate theory of human problem-solving if we were to understand decisions. Allen Newell, whom I had met at the Rand Corporation in 1952, held similar views. About 1954, he and I conceived the idea that the right way to study problem-solving was to simulate it with computer programs. Gradually, computer simulation of human cognition became my central research interest, an interest that has continued to be absorbing up to the present time.

One may object that I exaggerate the artificiality of our world. I shall plead guilty of overstatement, while protesting that the exaggeration is slight.

The world we live in today is much more a man-made, or artificial, world than it is a natural world. Almost every element in our environment shows evidence of human artifice. The temperature in which we spend most of our hours is kept artificially at 20 degrees Celsius; the humidity is added to or taken from the air we breathe; and the impurities we inhale are largely produced (and filtered) by man.

AM: Would you view your own decisions as something strictly ‘mechanical’?

HS: One of my few important decisions, and the best, was to persuade Dorothea Pye to marry me on Christmas Day, 1937. We have shared also the pleasures and responsibilities of raising three children, none of whom seem imitative of their parents’ professional directions, but all of whom have shaped for themselves interesting and challenging lives.

AM: You predicted that a computer chess program would beat the world chess-champion in ten years. That happened, but three decades late. What happened?

HS: I was a little too far-sighted with chess, but there was no way to do it with machines that were as slow as the ones back then.

In the middle of a game, when many pieces remain in play, each player typically has 30 or 40 moves. So after one move by each player (that’s called two "plies," or one "move") the board could show about 1,000 positions. By another complete move, there would be 1 million, and by the third move, 1 billion.

remote_rover
Remote robotic explorers, like the 1997 Mars Pathfinder mission, require a level of autonomous decision making
Credit: NASA/JPL

That kind of "combinatorial explosion" lead to this phenomenal analysis: that the number of possible unique chess games equals 10 to the power 120.

AM: So was IBM’s Deep Blue chess player ‘intelligent’, simply by its result of beating the world champion?

HS: AI folks use two definitions for intelligence: "What are the tasks, which when done by humans, lead us to impute intelligence?" and "What are the processes humans use to act intelligently?"

Measured against the first definition, Deep Blue certainly is intelligent. It partly qualifies, according to the second. It certainly did use an enormous amount of [computer] cycles, [a hallmark of brute force], but it also used a limited amount of rules.

These programs may contain thousands of "if-then" statements. This is a human-like strategy, because a large part of human knowledge is stored in this form. If our knowledge base could be measured, it would come to maybe 10,000 or 100,000 statements.

Thus a typical common-sense life rule might read, "If you are driving a car, do not aim at stationary objects like trees."

AM: A thinking machine consists of many ‘if-then’ rules, but how does one formulate these rules without explicit programming. In other words, not rote, but true learning?

HS: There are billions of neurons, and even more connections between neurons in the brain. That sounds complex. So the goal here is the goal of all science: You take something which you don’t understand — it looks extremely complex, it looks like a mystery — and you say: ‘There has to be order in this, there has to be a system in this, or it wouldn’t work. Let’s find out what it is, and when we understand something about the order, then we’ll see it much simpler.’

Not less impressive, not less effective, but simply more understandable. That is of course what we aim at in all this research.

AM: So you seem to remain very curious about the world. With the self-knowledge that a machine can simulate your patterns?

HS: I realized that you could formulate theories about human and social phenomena in language and pictures and whatever you wanted on the computer. The aesthetics of natural science and mathematics is at one with the aesthetics of music and painting–both inhere in the discovery of a partially concealed pattern.

Wonderful but not incomprehensible. This is the task of natural science: to show that the wonderful is not incomprehensible, to show how it can be comprehended–but not to destroy wonder.

The phrase ‘artificial intelligence’, was coined, I think right on the Charles River, at MIT. Our own research group at Rand and Carnegie Mellon University have preferred phrases like ‘complex information processing’ and ‘simulation of cognitive processes’.

At any rate, ‘artificial intelligence’ seems to be here to stay. In time it will become sufficiently idiomatic that it will not longer be the target of cheap rhetoric.

AM: Your own life has had some important ‘if-then’ rules. Would you have been attracted to the field of artificial intelligence if your own father hadn’t been an electrical engineer?

HS: My career was settled at least as much by drift as by choice.

My father had come to the United States in 1903 after earning his engineering diploma at the Technische Hochschule of Darmstadt, Germany. He was an inventor and designer of electrical control gear, later also a patent attorney. My mother, an accomplished pianist, was a third generation American, her forebears having been ’48ers who immigrated from Prague and Köln. Among my European ancestors were piano builders, goldsmiths, and vintners but to the best of my knowledge, no professionals of any kind.

Our dinner table at home was a place for discussion and debate.


Related Web Pages

Defining Life
Life from Scratch?
Viking and the Biology Experiments
Entropy and Evolution
What is Life?
Miller-Urey Experiment: Amino Acids from Scratch
The Envelope of Life? Please