Showing posts with label John McCarthy. Show all posts
Showing posts with label John McCarthy. Show all posts

Artificial Intelligence - Who Was John McCarthy?


John McCarthy  (1927–2011) was an American computer scientist and mathematician who was best known for helping to develop the subject of artificial intelligence in the late 1950s and pushing the use of formal logic in AI research.

McCarthy was a creative thinker who earned multiple accolades for his contributions to programming languages and operating systems research.

Throughout McCarthy's life, however, artificial intelligence and "formalizing common sense" remained his primary research interest (McCarthy 1990).

As a graduate student, McCarthy first met the concepts that would lead him to AI at the Hixon conference on "Cerebral Mechanisms in Behavior" in 1948.

The symposium was place at the California Institute of Technology, where McCarthy had just finished his undergraduate studies and was now enrolled in a graduate mathematics program.

In the United States, machine intelligence had become a subject of substantial academic interest under the wide term of cybernetics by 1948, and many renowned cyberneticists, notably Princeton mathematician John von Neumann, were in attendance at the symposium.

McCarthy moved to Princeton's mathematics department a year later, when he discussed some early ideas inspired by the symposium with von Neumann.

McCarthy never published the work, despite von Neumann's urging, since he believed cybernetics could not solve his problems concerning human knowing.

McCarthy finished a PhD on partial differential equations at Princeton.

He stayed at Princeton as an instructor after graduating in 1951, and in the summer of 1952, he had the chance to work at Bell Labs with cyberneticist and inventor of information theory Claude Shannon, whom he persuaded to collaborate on an edited collection of writings on machine intelligence.

Automata Studies received contributions from a variety of fields, ranging from pure mathematics to neuroscience.

McCarthy, on the other hand, felt that the published studies did not devote enough attention to the important subject of how to develop intelligent machines.

McCarthy joined the mathematics department at Stanford in 1953, but was fired two years later, maybe because he spent too much time thinking about intelligent computers and not enough time working on his mathematical studies, he speculated.

In 1955, he accepted a position at Dartmouth, just as IBM was preparing to establish the New England Computation Center at MIT.

The New England Computation Center gave Dartmouth access to an IBM computer that was installed at MIT and made accessible to a group of New England colleges.

McCarthy met IBM researcher Nathaniel Rochester via the IBM initiative, and he recruited McCarthy to IBM in the summer of 1955 to work with his research group.

McCarthy persuaded Rochester of the need for more research on machine intelligence, and he submitted a proposal to the Rockefeller Foundation for a "Summer Research Project on Artificial Intelligence" with Rochester, Shannon, and Marvin Minsky, a graduate student at Princeton, which included the first known use of the phrase "artificial intelligence." Despite the fact that the Dartmouth Project is usually regarded as a watershed moment in the development of AI, the conference did not go as McCarthy had envisioned.

The Rockefeller Foundation supported the proposal at half the proposed budget since it was for such an unique field of research with a relatively young professor as author, and because Shannon's reputation carried substantial weight with the Foundation.

Furthermore, since the event took place over many weeks in the summer of 1955, only a handful of the guests were able to attend the whole period.

As a consequence, the Dartmouth conference was a fluid affair with an ever-changing and unpredictably diverse guest list.

Despite its chaotic implementation, the meeting was crucial in establishing AI as a distinct area of research.

McCarthy won a Sloan grant to spend a year at MIT, closer to IBM's New England Computation Center, while still at Dartmouth in 1957.

McCarthy was given a post in the Electrical Engineering department at MIT in 1958, which he accepted.

Later, he was joined by Minsky, who worked in the mathematics department.

McCarthy and Minsky suggested the construction of an official AI laboratory to Jerome Wiesner, head of MIT's Research Laboratory of Electronics, in 1958.

McCarthy and Minsky agreed on the condition that Wiesner let six freshly accepted graduate students into the laboratory, and the "artificial intelligence project" started teaching its first generation of students.

McCarthy released his first article on artificial intelligence in the same year.

In his book "Programs with Common Sense," he described a computer system he named the Advice Taker that would be capable of accepting and understanding instructions in ordinary natural language from nonexpert users.

McCarthy would later define Advice Taker as the start of a study program aimed at "formalizing common sense." McCarthy felt that everyday common sense notions, such as comprehending that if you don't know a phone number, you'll need to look it up before calling, might be written as mathematical equations and fed into a computer, enabling the machine to come to the same conclusions as humans.

Such formalization of common knowledge, McCarthy felt, was the key to artificial intelligence.

McCarthy's presentation, which was presented at the United Kingdom's National Physical Laboratory's "Symposium on Mechansation of Thought Processes," helped establish the symbolic program of AI research.

McCarthy's research was focused on AI by the late 1950s, although he was also involved in a range of other computing-related topics.

In 1957, he was assigned to a group of the Association for Computing Machinery charged with developing the ALGOL programming language, which would go on to become the de facto language for academic research for the next several decades.

He created the LISP programming language for AI research in 1958, and its successors are widely used in business and academia today.

McCarthy contributed to computer operating system research via the construction of time sharing systems, in addition to his work on programming languages.

Early computers were large and costly, and they could only be operated by one person at a time.

McCarthy identified the necessity for several users throughout a major institution, such as a university or hospital, to be able to use the organization's computer systems concurrently via computer terminals in their offices from his first interaction with computers in 1955 at IBM.

McCarthy pushed for study on similar systems at MIT, serving on a university committee that looked into the issue and ultimately assisting in the development of MIT's Compatible Time-Sharing System (CTSS).

Although McCarthy left MIT before the CTSS work was completed, his advocacy with J.C.R.

Licklider, future office head at the Advanced Research Projects Agency, the predecessor to DARPA, while a consultant at Bolt Beranek and Newman in Cambridge, was instrumental in helping MIT secure significant federal support for computing research.

McCarthy was recruited to join what would become the second department of computer science in the United States, after Purdue's, by Stanford Professor George Forsythe in 1962.

McCarthy insisted on going only as a full professor, which he believed would be too much for Forsythe to handle as a young researcher.

Forsythe was able to persuade Stanford to grant McCarthy a full chair, and he moved to Stanford in 1965 to establish the Stanford AI laboratory.

Until his retirement in 2000, McCarthy oversaw research at Stanford on AI topics such as robotics, expert systems, and chess.

McCarthy was up in a family where both parents were ardent members of the Communist Party, and he had a lifetime interest in Russian events.

He maintained numerous professional relationships with Soviet cybernetics and AI experts, traveling and lecturing there in the mid-1960s, and even arranged a chess match between a Stanford chess computer and a Russian equivalent in 1965, which the Russian program won.

He developed many foundational concepts in symbolic AI theory while at Stanford, such as circumscription, which expresses the idea that a computer must be allowed to make reasonable assumptions about problems presented to it; otherwise, even simple scenarios would have to be specified in such exacting logical detail that the task would be all but impossible.

McCarthy's accomplishments have been acknowledged with various prizes, including the 1971 Turing Award, the 1988 Kyoto Prize, admission into the National Academy of Sciences in 1989, the 1990 Presidential Medal of Science, and the 2003 Benjamin Franklin Medal.

McCarthy was a brilliant thinker who continuously imagined new technologies, such as a space elevator for economically transporting stuff into orbit and a system of carts strung from wires to better urban transportation.

In a 2008 interview, McCarthy was asked what he felt the most significant topics in computing now were, and he answered without hesitation, "Formalizing common sense," the same endeavor that had inspired him from the start.

~ Jai Krishna Ponnappan

Find Jai on Twitter | LinkedIn | Instagram

You may also want to read more about Artificial Intelligence here.

See also: 

Cybernetics and AI; Expert Systems; Symbolic Logic.

References & Further Reading:

Hayes, Patrick J., and Leora Morgenstern. 2007. “On John McCarthy’s 80th Birthday, in Honor of His Contributions.” AI Magazine 28, no. 4 (Winter): 93–102.

McCarthy, John. 1990. Formalizing Common Sense: Papers, edited by Vladimir Lifschitz. Norwood, NJ: Albex.

Morgenstern, Leora, and Sheila A. McIlraith. 2011. “John McCarthy’s Legacy.” Artificial Intelligence 175, no. 1 (January): 1–24.

Nilsson, Nils J. 2012. “John McCarthy: A Biographical Memoir.” Biographical Memoirs of the National Academy of Sciences.

Artificial Intelligence - The Frame Problem.


 John McCarthy and Patrick Hayes discovered the frame problem in 1969.

The problem concerns representing the effects of actions in logic-based artificial intelligence.

Formal logic is used to define facts about the world, such as a car can be started when the key is placed in the ignition and turned and that pressing the accelerator causes it to move forward.

However, the latter fact does not explicitly state that the car remains on after pressing the accelerator.

To correct this, the fact must be expanded to “pressing the accelerator moves the car forward and does not turn it off.” However, this fact must be augmented further to describe many other scenarios (e.g., that the driver also remains in the vehicle) (e.g., that the driver also remains in the vehicle).

The frame problem highlights an issue in logic involving the construction of facts that do not require enumerating thousands of trivial effects.

After its discovery by artificial intelligence researchers, the frame issue was taken up by philosophers.

Their version of the issue could be appropriately dubbed the world update problem since it involves updating frames of reference.

For example, how do you know your dog (or other pet) is where you last saw them without seeing them again? In a philosophic perspective, the frame issue is concerned with how well a person's perception of their environment corresponds to reality and when that understanding should be changed.

As intelligent agents plan activities in more complicated environments, they will have to deal with this issue.

To solve the logic version of the frame problem, a number of solutions have been proposed.

The philosophical problem, on the other hand, remains unsolved.

Both must be solved in order for artificial intelligence to behave intelligently.

Jai Krishna Ponnappan

You may also want to read more about Artificial Intelligence here.

See also: 

McCarthy, John.

Further Reading:

McCarthy, John, and Patrick J. Hayes. 1969. “Some Philosophical Problems from the Standpoint of Artificial Intelligence.” In Machine Intelligence, vol. 4, edited by Donald Michie and Bernard Meltzer, 463–502. Edinburgh, UK: Edinburgh University Press.

Shanahan, Murray. 1997. Solving the Frame Problem: A Mathematical Investigation of  the Common Sense Law of Inertia. Cambridge, MA: MIT Press.

Shanahan, Murray. 2016. “The Frame Problem.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta.

Artificial Intelligence - What Is The Dartmouth AI Conference?


    The Dartmouth Conference on Artificial Intelligence, officially known as the "Dartmouth Summer Research Project on Artificial Intelligence," was held in 1956 and is frequently referred to as the AI Constitution.

    • The multidisciplinary conference, held on the Dartmouth College campus in Hanover, New Hampshire, brought together specialists in cybernetics, automata and information theory, operations research, and game theory.
    • Claude Shannon (the "father of information theory"), Marvin Minsky, John McCarthy, Herbert Simon, Allen Newell ("founding fathers of artificial intelligence"), and Nathaniel Rochester (architect of IBM's first commercial scientific mainframe computer) were among the more than twenty attendees.
    • Participants came from the MIT Lincoln Laboratory, Bell Laboratories, and the RAND Systems Research Laboratory.

    The Rockefeller Foundation provided a substantial portion of the funding for the Dartmouth Conference.

    The Dartmouth Conference, which lasted around two months, was envisaged by the organizers as a method to make quick progress on computer models of human cognition.

    • "Every facet of learning or any other trait of intelligence may in theory be so clearly characterized that a computer can be constructed to replicate it," organizers said as a starting point for their deliberations (McCarthy 1955, 2).

    • In his Rockefeller Foundation proposal a year before to the summer meeting, mathematician and principal organizer John McCarthy created the phrase "artificial intelligence." McCarthy subsequently said that the new name was intended to establish a barrier between his study and the discipline of cybernetics.
    • He was a driving force behind the development of symbol processing techniques to artificial intelligence, which were at the time in the minority.
    • In the 1950s, analog cybernetic techniques and neural networks were the most common brain modeling methodologies.

    Issues Covered At The Conference.

    The Dartmouth Conference included a broad variety of issues, from complexity theory and neuron nets to creative thinking and unpredictability.

    • The conference is notable for being the site of the first public demonstration of Newell, Simon, and Clifford Shaw's Logic Theorist, a program that could independently verify theorems stated in Bertrand Russell and Alfred North Whitehead's Principia Mathematica.
    • The only program at the conference that tried to imitate the logical features of human intellect was Logic Theorist.
    • Attendees predicted that by 1970, digital computers would have become chess grandmasters, discovered new and important mathematical theorems, produced passable language translations and understood spoken language, and composed classical music.
    • Because the Rockefeller Foundation never received a formal report on the conference, the majority of information on the events comes from memories, handwritten notes, and a few papers authored by participants and published elsewhere.

    Mechanization of Thought Processes

    Following the Dartmouth Conference, the British National Physical Laboratory (NPL) hosted an international conference on "Mechanization of Thought Processes" in 1958.

    • Several Dartmouth Conference attendees, including Minsky and McCarthy, spoke at the NPL conference.
    • Minsky mentioned the Dartmouth Conference's relevance in the creation of his heuristic software for solving plane geometry issues and the switch from analog feedback, neural networks, and brain modeling to symbolic AI techniques at the NPL conference.
    • Neural networks did not resurface as a research topic until the mid-1980s.

    Dartmouth Summer Research Project 

    The Dartmouth Summer Research Project on Artificial Intelligence was a watershed moment in the development of AI. 

    The Dartmouth Summer Research Project on Artificial Intelligence, which began in 1956, brought together a small group of scientists to kick off this area of study. 

    To mark the occasion, more than 100 researchers and academics gathered at Dartmouth for AI@50, a conference that celebrated the past, appraised current achievements, and helped seed ideas for future artificial intelligence research. 

    John McCarthy, then a mathematics professor at the College, convened the first gathering. 

    The meeting would "continue on the basis of the premise that any facet of learning or any other attribute of intelligence may in theory be so clearly characterized that a computer can be constructed to replicate it," according to his plan. 

    The director of AI@50, Professor of Philosophy James Moor, explains that the researchers who came to Hanover 50 years ago were thinking about methods to make robots more aware and sought to set out a framework to better comprehend human intelligence.

    Context Of The Dartmouth AI Conference:

    Cybernetics, automata theory, and sophisticated information processing were all terms used in the early 50s to describe the science of "thinking machines." 

    The wide range of names reflects the wide range of intellectual approaches. 

    In, John McCarthy, a Dartmouth College Assistant Professor of Mathematics, wanted to form a group to clarify and develop ideas regarding thinking machines. 

    • For the new field, he chose the moniker 'Artificial Intelligence.' He picked the term mainly to escape a concentration on limited automata theory and cybernetics, which was largely focused on analog feedback, as well as the possibility of having to accept or dispute with the aggressive Norbert Wiener as guru. 
    • McCarthy addressed the Rockefeller Foundation in early to seek money for a summer seminar at Dartmouth that would attract roughly 150 people. 
    • In June, he and Claude Shannon, then at Bell Labs, met with Robert Morison, Director of Biological and Medical Research, to explore the concept and potential financing, but Morison was skeptical if money would be made available for such a bold initiative. 

    McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon officially proposed the proposal in September. The term "artificial intelligence" was coined as a result of this suggestion. 

    According to the proposal, 

    • We suggest that during the summer of at Dartmouth College in Hanover, New Hampshire, a -month, -man artificial intelligence research be conducted. 
    • The research will be based on the hypothesis that any part of learning, or any other characteristic of intelligence, can be characterized exactly enough for a computer to imitate it. 
    • It will be attempted to figure out how to get robots to speak, develop abstractions and ideas, solve issues that are now reserved for people, and improve themselves. 
    • We believe that if a properly chosen group of scientists worked on one or more of these topics together for a summer, considerable progress might be accomplished. 
    • Computers, natural language processing, neural networks, theory of computing, abstraction, and creativity are all discussed further in the proposal (these areas within the field of artificial intelligence are considered still relevant to the work of the field). 

    He remarked, "We'll focus on the difficulty of figuring out how to program a calculator to construct notions and generalizations. 

    Of course, this is subject to change once the group meets." Ray Solomonoff, Oliver Selfridge, Trenchard More, Arthur Samuel, Herbert A. Simon, and Allen Newell were among the participants at the meeting, according to Stottler Henke Associates. 

    The real participants arrived at various times, most of which were for far shorter periods of time. 

    • Rochester was replaced for three weeks by Trenchard More, and MacKay and Holland were unable to attend—but the project was prepared to commence. 
    • Around June of that year, the first participants (perhaps simply Ray Solomonoff, maybe with Tom Etter) came to Dartmouth College in Hanover, New Hampshire, to join John McCarthy, who had already set up residence there. 
    • Ray and Marvin remained at the Professors' apartments, while the most of the guests stayed at the Hanover Inn.

    List Of Dartmouth AI Conference Attendees:

    1. Ray Solomonoff
    2. Marvin Minsky
    3. John McCarthy
    4. Claude Shannon
    5. Trenchard More
    6. Nat Rochester
    7. Oliver Selfridge
    8. Julian Bigelow
    9. W. Ross Ashby
    10. W.S. McCulloch
    11. Abraham Robinson
    12. Tom Etter
    13. John Nash
    14. David Sayre
    15. Arthur Samuel
    16. Kenneth R. Shoulders
    17. Shoulders' friend
    18. Alex Bernstein
    19. Herbert Simon
    20. Allen Newell

    ~ Jai Krishna Ponnappan

    You may also want to read more about Artificial Intelligence here.

    See also: 

    Cybernetics and AI; Macy Conferences; McCarthy, John; Minsky, Marvin; Newell, Allen; Simon, Herbert A.

    References & Further Reading:

    Crevier, Daniel. 1993. AI: The Tumultuous History of the Search for Artificial Intelligence. New York: Basic Books.

    Gardner, Howard. 1985. The Mind’s New Science: A History of the Cognitive Revolution. New York: Basic Books.

    Kline, Ronald. 2011. “Cybernetics, Automata Studies, and the Dartmouth Conference on Artificial Intelligence.” IEEE Annals of the History of Computing 33, no. 4 (April): 5–16.

    McCarthy, John. 1955. “A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence.” Rockefeller Foundation application, unpublished.

    Moor, James. 2006. “The Dartmouth College Artificial Intelligence Conference: The Next Fifty Years.” AI Magazine 27, no. 4 (Winter): 87–91.

    Solomonoff, R.J.The Time Scale of Artificial Intelligence; Reflections on Social Effects, Human Systems Management, Vol 5 1985, Pp 149-153

    Moor, J., The Dartmouth College Artificial Intelligence Conference: The Next Fifty years, AI Magazine, Vol 27, No., 4, Pp. 87-9, 2006

    ump up to:

    Kline, Ronald R., Cybernetics, Automata Studies and the Dartmouth Conference on Artificial Intelligence, IEEE Annals of the History of Computing, October–December, 2011, IEEE Computer Society

    McCorduck, P., Machines Who Think, A.K. Peters, Ltd, Second Edition, 2004

    Nilsson, N., The Quest for Artificial Intelligence, Cambridge University Press, 2010

    Kline, Ronald R., Cybernetics, Automata Studies and the Dartmouth Conference on Artificial Intelligence, IEEE Annals of the History of Computing, October–December, 2011, IEEE Computer Society, (citing letters, from Rockefeller Foundation Archives, Dartmouth file6, 17, 1955 etc.

    McCarthy, J., Minsky, M., Rochester, N., Shannon, C.E., A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence., August, 1955

    McCarthy, John; Minsky, Marvin; Rochester, Nathan; Shannon, Claude (1955), A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence, archived from the original on 2007-08-26, retrieved 2006-04-09 retrieved 10:47 (UTC), 9th of April 2006

     Stottler-Henke retrieved 18:19 (UTC), 27th of July 2006

    Nilsson, N., The Quest for Artificial Intelligence, Cambridge University Press, 2010, P. 53

    Solomonoff, R.J., dart56ray622716talk710.pdf, 1956 URL:{

    McCarthy, J., List, Sept., 1956; List among Solomonoff papers to be posted on website 1956

    Nilsson, N., The Quest for Artificial Intelligence, Cambridge University Press, 2010,
    personal communication

    McCorduck, P., Machines Who Think, A.K. Peters, Ltd, Second Edition, 2004.

    What Is Artificial General Intelligence?

    Artificial General Intelligence (AGI) is defined as the software representation of generalized human cognitive capacities that enables the ...