Quotations

(a subtopic of the Reference Shelf)

Amarel, Saul -- Many problems of practical importance are problems of reasoning about actions. In these problems, a course of action has to be found that satisfies a number of specified conditions. Everyday examples of reasoning about actions include planning an airplane trip, organizing a dinner party, etc. ... A problem of reasoning about actions (Simon, 1966) is given in terms of an initial situation, a terminal situation, a set of feasible actions, and a set of constraints that restrict the applicability of actions. The task of a problem solver is to find the best sequence of permissable actions that can transform the initial situation into the terminal situation. From On Representations of Problems of Reasoning About Actions. Printed in Readings in Artificial Intelligence, ed. by Bonnie Lynn Webber and Nils J. Nilsson, 1981. Palo Alto, CA: Tioga Pub. Co.

Aristotle -- If every tool, when ordered, or even of its own accord, could do the work that befits it, just as the creations of Daedalus moved of themselves . . . If the weavers' shuttles were to weave of themselves, then there would be no need either of apprentices for the master workers or of slaves for the lords. From Atheniensium Republica.

Aristotle -- Speech is the representation of the mind, and writing is the representation of speech. From De Interpretatione I.

Babbage, Charles -- The whole of the developments and operations of analysis are now capable of being executed by machinery. . . . As soon as an Analytical Engine exists, it will necessarily guide the future course of science. From Passages from the Life of a Philosopher, 1832.

Babbage, Charles -- . . .every game of skill is susceptible of being played by an automaton. From The Life of a Philosopher.

Bailey, James -- Humanity is not at stake in this. Old ideas, old oversimplified ideas, are at stake, but that's good. Humans and computers are going to make a good team. From "Our Machines, Ourselves," Harper's Magazine, May 1997.

Barrow, Harry G. and J. M. Tenenbaum -- There are fundamental questions to be answered about the architecture of a visual system. For nearly two decades, the field has assumed that the visual system can be decomposed into independent modules, each performing a well-defined function, like estimating color, and that their outputs are integrated at a later stage. Is this a valid hypothesis? From Retrospective on "Interpreting Line Drawings as Three-Dimensional Surfaces." 1993. Artificial Intelligence 59(1/2): 71-80.

Buchanan, Bruce -- In calling AI an empirical science, we presuppose regularities in intelligent behavior of people and computers that can be discovered by observation and experimentation. From Artificial Intelligence as an Experimental Science. Reprinted in Aspects of Artificial Intelligence, ed. by James H. Fetzer, 1988, Kluwer Academic Publishers.

Burton, Richard R. and John Seely Brown -- For an informal environment to be fully effective as a learning activity, it often must be augmented by tutorial guidance that recognizes and explains weaknesses in the student's decisions or suggests ideas when the student appears to have none. This is a significant challenge requiring many of the skills analogous to those of a coach or laboratory instructor. The tutor or coach must be perceptive enough to make relevant comments but not so intrusive as to destroy the fun inherent in the game. From An Investigation of Computer Coaching for Informal Learning Activities. Printed in Intelligent Tutoring Systems, ed. by D. Sleeman and J. S. Brown, 1982. London/New York: Academic Press.

Dijkstra, E. W. -- LISP has jokingly been called "the most intelligent way to misuse a computer." I think that description is a great compliment because it transmits the full flavor of liberation: it has assisted a number of our most gifted fellow humans in thinking previously impossible thoughts. From the Humble Programmer, Communications of the ACM 15(10) - 1972.

• Also see: Tributes

Dyson, Esther -- Some of the most successful applications of AI are those inwhich the artificial intelligence is spread like raisins in a loaf ofraisin bread: the raisins do not occupy much space, but theyoften provide the principal source of nutrition.

• Also see:
  • Video: ArsDigita University Curriculum - Artificial Intelligence course taught by Patrick Winston. Lecture #1 (of 4): AI Overview, Rule-Based Expert Systems and Knowledge Engineering: at 34:04.
  • Speakers discuss past, present and future of AI. ByDenise Brehm. MIT News Office(October 1, 1997). "In describing some of the lessons learned by researchers since the eighties, he quoted Esther Dyson, editor of the influential trade publication Release 1.0, who predicted that AI would not become commercially significant until it became strategically important, 'like raisins embedded in raisin bread.' Professor Winston went on to describe how AI was doing just that."

Epstein, Susan L. -- A person who has played other games approaches a newly-introduced game with certain expectations: there will be rules, players, playing pieces, perhaps a board, and the players will take turns. Although he may be new to the game, experience has taught him what there is to be learned about a game and how to learn it. . . .The AI approach to game playing thus far has been to build a program that plays only a single game, and plays it very well. . . .People learn to play well. In contrast, many game playing machines have little ability to improve without programmer intervention. . . .As they pore over Hitech's experiences [in chess], it is the researchers who are learning from Hitech's experiences, not the machine. From The Intelligent Novice: Learning to Play Better, in the book Heuristic Programming in Artificial Intelligence: The First Olympiad, edited by David Levy, 1989. Chichester, UK: Ellis Horwood.

Feigenbaum, Edward; McCorduck, Pamela; and Nii, H. Penny -- Consider how much more valuable than data is the company's knowledge. In some cases it's unique expertise. Will the standard methods for protection suffice? . . . Who owns the knowledge, anyway?. . .Who gets to hold the copyright on an expert's lifetime of experience in performing his niche task? From The Rise of the Expert Company, 1988. New York: Times Books/Random Hous, Inc.

Feigenbaum, Edward; McCorduck, Pamela; and Nii, H. Penny -- Today's expert systems deal with domains of narrow specialization. . .For expert systems to perform competently over a broad range of tasks, they will have to be given very much more knowledge. ... The next generation of expert systems ... will require large knowledge bases. How will we get them? From The Rise of the Expert Company, 1988. New York: Times Books/Random House, Inc.

Feigenbaum, Edward; McCorduck, Pamela; and Nii, H. Penny -- The user of the library of the future need not be a person. It may be another knowledge system -- that is, any intelligent agent with a need for knowledge. Such a library will be a network of knowledge systems, in which people and machines collaborate. 1988. From The Rise of the Expert Company, p. 257. New York: Times Books/Random House, Inc.

Grosz, Barbara -- Progress on building computer systems that process natural language in any meaningful sense (i.e., systems that interact reasonably with humans in natural language) requires considering language as a part of a larger communicative situation. In this larger situation, the participants in a conversation and their states of mind are as important to the interpretaion of an utterance as the linguistic expression from which it is formed. ... regarding language as communication requires consideration of what is said (literally), what is intended, and the rrelationship between the two. --From Utternace and Objective: Issues in Natural Language Communication, IJCAI-79, Vol. 2, p. 1067.

Grosz, Barbara, and Randall Davis -- To achieve their full impact, computer systems must have more than processing power--they must have intelligence. They need to be able to assimilate and use large bodies of information and collaborate with and help people find new ways of working together effectively. The technology must become more responsive to human needs and styles of work, and must employ more natural means of communication. From A Report to ARPA on Twenty-First Century Intelligent Systems. AI Magazine 15 (3): 10-20.

Huff, K. and Selfridge, O. -- For systems of the future, we need to think in terms of shifting the burden of evolution from programmers to the systems themselves. ... to build systems that can take some responsibility for their own evolution. From Evolution in Future Intelligent Information Systems. In Proceedings of the International Workshop on the Development of Intelligent Information Systems.

Landauer, Thomas K. -- The story of the human race is one of ever-increasing intellectual capability. Since our early cave-dwelling ancestors, our brains have gotten no bigger, our hands no more nimble, but there has been a steady accretion of tools for intellectual work--how to grow crops, domesticate animals, build shelters, paint paintings.It includes governing and inspiring and, unfortunately, waging wars. It includes how to build and operate airlines, television sets, and football teams. This shared capacity was first manifest in language, later in writing, math and science, and in the huge collections of experience and discovery stored in books and libraries. By comparison with our forebears, each of us has become a genius. From The Trouble With Computers, 1995, MIT Press. Page 365.

Leonardo da Vinci -- The science of painting deals with all the colours of the surfaces of bodies and with the shapes of the bodies thus enclosed; with their relative nearness and distance; with the degrees of diminutions required as distances gradually increase; moreover, this science is the mother of perspective, that is, of the science of visual rays. From Paragone.

Levy, David N. L. -- Anyone who has read a number of bridge columns in newspapers will be struck by the frequency with which even the world's top players make mistakes. . . .The key . . .lies in writing a program that can play the cards "perfectly." From The Million Pound Bridge Program by David Levy, reprinted in Heuristic Programming in Artificial Intelligence: The First Olympiad, ed. by David Levy, 1989. Chichester, UK: Ellis Horwood.

McCarthy, John -- Artificial intelligence cannot avoid philosophy. If a computer program is to behave intelligently in the real world, it must be provided with some kind of framework into which to fit particular facts it is told or discovers. This amounts to at least a fragment of philosophy, however naive. From "Mathematical Logic in Artificial Intelligence." Reprinted in Artificial Intelligence Debate, ed. Stephen Graubard, 1990, MIT Press.

McCorduck, Pamela -- Perhaps the earliest examples of the urge to make artificial persons are the Greek Gods, those wonderful superhumans who seem to behave as we would if only we had the means. . . . As a present from Zeus to Europa, Hephaestus makes Talos, a man of bronze whose duty is to patrol the beaches of Crete three times a day. He thwarts invadersby hurling great rocks at them, or by heating himself red hot and squeezing trespassers in a warm embrace. From Machines Who Think, p. 4, 1979, W. H. Freeman and Co.

McDermott, John -- To be useful, a system has to do more than just correctly perform some task. From R1 at the Age of 12: Lessons from an Elementary School Achiever. 1993. Printed in Artificial Intelligence 59(1/2): 241 - 247.

Michie, Donald -- Why do we not, since the phenomena are well known, build a "knowledge refinery" as the basis of a new industry, comparable in some ways to the industry of petroleum refining, only more important in the long run? The product to be refined is codified human knowledge. From A Prototype Knowledge Refinery, printed in Inteligent Systems: The Unprecedented Opportunity, ed. by J.E. Hayes and D, Michie. Chichester, UK: Ellis Horwood (1983).

Minsky, Marvin -- In the 1960s and 1970s, students frequently asked, "Which kind of representation is best?" and I usually replied that we'd need more research. . .But now I would reply: To solve really hard problems, we'll have to use several different representations. This is because each particular kind of data structure has its own virtues and deficiencies, and none by itself would seem adequate for all the different functions involved with what we call common sense. From Logical vs. Analogical . . .AI Magazine 12.2

Minsky, Marvin -- This book assumes that any brain, machine, or other thing that has a mind must be composed of smaller things that cannot think at all. ... Are minds machines? Of that I've raised no doubt at all but have only asked, what kind of machines? From The Society of Mind, p. 322. 1985. New York: Simon & Schuster.

Minsky, Marvin -- It is not that the games and mathematical problems are chosen because they are clear and simple; rather it is that they give us, for the smallest initial structures, the greatest complexity, so that one can engage some really formidable situations after a relatively minimal diversion into programming. From Semantic Information Processing, p. 12. Cambridge, MA: MIT Press (1968).

Minsky, Marvin -- Any powerful heuristic program is bound to contain a variety of different methods and techniques. At each step of the problem-solving process the machine will have to decide what aspect of the problem to work on, and then which method to use. A choice must be made, for we usually cannot afford to try all the possibilities. From Steps Toward Artificial Intelligence. Reprinted in Computers and Thought, ed. by Edward A. Feigenbaum and Julian Feldman, re-issued 1995. Cambridge, MA: MIT Press.

Minsky, Marvin -- if we understood something just one way, we would not understand it at all. That is why the seekers of the "real" meanings never find them. This holds true especially for words like 'understand'. ''From Music, Mind, and Meaning (1981).

Moravec, Hans -- Domestic machines such as food processors, vacuum cleaners, and microwave ovens do not fill the void in families where all adults work outside the home. The need has existed for many decades: When will there be a robot to help around the house? From Mind Children, 1988, Harvard University Press.

Newell, Allen -- Exactly what the computer provides is the ability not to be rigid and unthinking but, rather, to behave conditionally. That is what it means to apply knowledge to action: It means to let the action taken reflect knowledge of the situation, to be sometimes this way, sometimes that, as appropriate. ... In sum, technology can be controlled especially if it is saturated with intelligence to watch over how it goes, to keep accounts, to prevent errors, and to provide wisdom to each decision. From Fairytales, AI Magzine 13.4 (1992).

Newell, Allen -- From where I stand, it is easy to see the science lurking in robotics. It lies in the welding of intelligence to energy. That is, it lies in intelligent perception and intelligent control of motion. From The Scientific Relevance of Robotics (Remarks at the Dedication of the CMU Robotics Institute). AI Magazine 2(1): 24-26, 34 (Winter 1980).

Nilsson, Nils -- The real trick in designing an efficient automatic problem solver is to search at the highest level permitted by the available information about the problem and about how it might be solved. From Problem-Solving Methods in Artificial Intelligence, 1971. New York: McGraw Hill Book Company.

Nilsson, Nils -- The big problem for AI is what to say not how to say it. The predicate calculus[formal logic] does no more than provide a uniform language in which knowledge about the world (after we have it!) can be expressed and reasoned about. From Artificial Intelligence: A New Synthesis. San Francisco, CA: Morgan Kaufmann Publishers (1998).

Norman, Donald A. -- This new field -- Human Factors in Computer Systems -- contains an unruly mixture of theoretical issues and practical problems. Just as it is important that our theoretical concerns have breadth, generality, and usability, so too is it important that we understand the practical problems.... The technology upon which the human-computer interface is built changes rapidly relative to the time with which psychological experimentation yields answers. Our design principles must be of sufficient generality to outlast the technological demands of the moment. From Design Principles for Human-Computer Interfaces. Reprinted in Readings in Human-Computer Interaction, ed. by Ronald M. Baecker and William A.S. Buxton. San Mateo, CA: Morgan Kaufmann Press (1987).

Petroski, Henry -- "Our expectations for a technology rise with its advancement." From page 83 of his book, The Evolution of Useful Things. New York: Vintage Books (1994).

Pohl, Fred -- A good science fiction story should be able to predict not the automobile but the traffic jam. Taken from the book Close Encounters? Science and Science Fiction, by Robert Lambourne, Michael Shallis and Michael Shortland, 1990. Bristol and New York: Adam Hilger. Pollack, Martha -- We want to build intelligent actors, not just intelligent thinkers. Indeed, it is not even clear how one could assess intelligence in a system that never acted--or, put otherwise, how a system could exhibit intelligence in the absense of action. From IJCAI-1991 Computers and Thought Lecture. Reprinted in Artificial Intelligence 57 (1992): 43-68.

Reddy, Raj -- Systems were built in the 1950s for vowel recognition and digit recognition, producing creditable performance. But these techniques ... could not be extrapolated to more sophisticated systems. This led to the appreciation that linguistic and contextual cues must be brought to bear on the recognition strategy...A native speaker uses, subconsciously, knowledge of the language, the environment, and the context in understanding a sentence. ... To approach human performance, a machine must also use all the available knowledge sources effectively. From Speech Recognition by Machine: A Review. Readings in Speech Recognition, edited by Alex Waibel and Kai-Fu Lee, 1990. San Mateo, CA: Morgan Kaufmann Press.

Rissland, Edwina -- Can we move interface design further along the spectrum from art to science? ...Many of the discussions about interfaces are similar to those about what is art, or what is good art. That is, a central issue is how to judge an interface and how to determine what makes one interface better than another. From Ingredients of Intelligent User Interfaces. Reprinted in Readings in Human-Computer Interaction, ed. by Ronald M. Baecker and William A.S. Buxton, 1987. Los Altos: Morgan Kaufmann Publishers.

Rosenbloom, Paul S. -- . . .When I began this effort, my knowledge of both Othello and game-playing techniques was rudimentary. . . .Since then, with about 5 man-months of effort, IAGO has been brought up to world-championship level. 1982. From A World-Championship-Level Othello Program, Artifical Intelligence 19, pp.279-320.

Samuel, Arthur L. -- Alpha-beta pruning can be explained simply as a technique for not exploring those branches of a search tree that analysis indicates not to be of further interest either to the player making the analysis (this is obvious) or to his opponent (and this is frequently overlooked.) From Machine Learning Using the Game of Checkers II, reprinted in Computer Games I ed. bu David L. Levy, 1988, New York: Springer Verlag.

Selfridge, Oliver G. -- Instead of worrying about whether a particular machine can be intelligent, it is far more important to make a piece of software that is intelligent. From The Gardens of Learning. AI Magazine 14(2) (Summer 1993): p. 37.

Selfridge, Oliver G. -- If an expert systems--brilliantly designed, engineered and implemented--cannot learn not to repeat its mistakes, it is not as intelligent as a worm or a sea anemone or a kitten. From The Gardens of Learning, AI Magazine 14(2) (Summer 1993): p. 41.

Shapiro, Stuart C. and Howard R. Smith -- When humans play the game, a large easily accessible vocabulary seems to be the most important determiner of victory. One might, therefore, think that it would be easy to write a program that plays the SCRABBLE Crossword Game at championship level. However, several issues are not so clear: How should the lexicon be organized for maximum usefulness? How should a program decide where to play? How can a program take advantage of the small literature on the strategy and tactics of the game? What is the relative importance of a good memory for words vs. skillful decisions about what letters to use or not to use and where to play? From Scrabble Crossword Game Playing Program, reprinted in Computer Games I, ed by David L. Levy, 1988, New York: Springer Verlag.

Simon, Herbert A. -- Conclusion: Computers Think--and Often Think Like People. The human mind does not reach its goals mysteriously or miraculously. Even its sudden insights and "ahas" are explainable in terms of recognition processes, well-informed search, knowledge-prepared experiences of surprise, and changes in representation motivated by shifts in attention. When we incorporate these processes into our theory, as empirical evidence says we should, the unexplainable is explained. From Machine As Mind, 1995.

Simon, Herbert A. -- Open up the box of a computer, and you won't find any numbers in there. You'll find electromagnetic fields. Just as if you open up a person's brain case, you won't find symbols; yo'u'l find neurons. You can use those things, either neurons or electromagnetic fields, to represent any patterns you like. A computer could care less whether those patterns denote words, numbers, or pictures. Sure, in one sense, there are bits inside a computer, but what's important is not that they can do fast arithmetic but that they can manipulate symbols. That's how humans can think, and that's the basic hypothesis I operate from. From OMNI Magazine Interview (June 1994)

Sycara, Katia -- A multi-agent system can be defined as a loosely coupled network of problem solvers that interact to solve problems that are beyond the individual capabilities or knowledge of each problem solver. From The Many Faces of Agents. AI Magazine 19 (2): Summer 1998.

Waibel, Alex and Kai-Fu Lee -- ...Simple inquiries about bank balance, movie schedules, and phone call transfers can already be handled by small-to-medium sized vocabulary, speaker independent, telephone-speech recognizers. Voice activated data entry is particulary useful in medical or darkroom applications, where hands and eyes are unavailable as normal input medium, or in hands-busy or eyes-busy command and control applications. Speech could be used to provide more accessibility for the handicapped (wheelchairs, robotic aids, etc.) and to create high-tech amenities (intelligent houses, cars, etc.) From Readings in Speech Recognition, 1990. San Mateo: Morgan Kaufmann Press.

Waltz, David L. -- . . . because computers lack bodies and life experiences comparable to humans, intelligent systems will probably be inherently different from humans. . . . If we are to build machines that are as intelligent as people, we have three problems to solve: we must establish a science of cognition; we must engineer the software, sensors, and effectors for a full system; and we must devise adequate hardware. From The Prospects for Building Truly Intelligent Machines, Reprinted in The Artificial Intelligence Debate, ed. Stephen Graubard, 1988, MIT Press.

Wilcox, Bruce -- Chess and Go are both simplified examples of "inexact problems" that have no clear solution algorithm. Humans can be quite skilled at these problems, even if it is not clear how they do it. . . . Go is a whole new challenge requiring insights into human thoughts and new programming techniques. Because global changes occur slowly in Go, the game is much better suited to studying complex information management and decision making than is chess. From Computer Go, reprinted in Computer Games II, ed. by David L. Levy, 1988. New York: Springer Verlag.

Also See:

• Artificial Intelligence Quotations from Stottler Henke Associates, Inc.