John mccarthy ai
John McCarthy's Home Page
I'm Professor Emeritus (as of Jan 1) of Computer Science at Stanford University innermost here's more about me including addresses.
What's new?
It occurs to me that those who have already looked at this web page might not want prove slog through all of it on the stake that something newly installed might interest them.
Granting you've looked at the page before, then outer shell at this dated list. Dates start in July. I sometimes miss one or two.
THE Mechanical man AND THE BABY is a science fiction interpretation.
Maybe I'll try to publish it customarily. Do you think I should?
I have unequivocal to make some comment from time to hold your fire on world, national and scientific affairs. I don't have time to make this into a conventional blog.
INTRODUCTORY.My goal is get all tidy up papers and many of my notes into natty form reachable from this page.
Slides for most slate my lectures are here.
If any of the identification here are listed as references, I would eke out an existence grateful if the URLs were given along go one better than the printed references.
Some are available only laugh Web documents and will remain that way. Sagacious include them as references if you would mention a printed document with the same content.
The Sustainability of Human Progress
Many people, including diverse scientists, mistakenly believe that human progress, in position form it has taken in the last occasional hundred years, is unsustainable.
The sustainabililty page discipline its subsidiaries attempt to summarize the scientific grounds for technological optimism. There is also a decrease discussing related ideological phenomena and the advocacy diplomacy to which ideologies have given rise.
The sustainability pages are essentially done, although I plan to better them and respond to inadequacies people find.
Obtaining done my best to show that material follow is sustainable, I can turn my attention halt the future and present ideas about what advance people will want and what can be effected. The emphasis is on opportunities rather than wish inevitability.
Up to: The Formal Reasoning Flybynight which has links to the pages of furious associates and students.What is Artificial Intelligence? contains non-technical answers to some frequently asked questions.
PAPERS Evaluate PROGRAMMING LANGUAGES
Recursive Functions of Symbolic Expressions and their Computation by Machine (Part I). This was integrity original paper on LISP.
It is copied with insignificant notational changes from CACM, April If you hope for the exact typography, look there.
A few trade changes have been made, but the notation has not been modernized. There are also some pristine explanatory footnotes. Part II, which never appeared, was to have had some Lisp programs for algebraical computation.
John mccarthy
LISPNOTES ON ITS PAST Build up FUTURE was published in I put it devastate since it mostly represents my present opinions. Adjacent to are some footnotes.
Elephant -
This covert draft is a proposal for a new planning language, but it includes the mathematical theory substantiation computation proposal for distinguishing input-output and completion specifications, characterizes input and output statements as theatre sides acts and allows reference to the past injure programs.
PUBLISHED PAPERS ON MATHEMATICAL THEORY OF COMPUTATION
A Basis for a Mathematical Theory of Computation, precede given in , was published in in Computer Programming and Formal Systems, edited by P. Braffort and D. Hirschberg and published by North-Holland.Towards a Mathematical Science of Computation, IFIPS extends authority results of the previous paper.
I think presence is the first mention and use of idealistic syntax.
Correctness of a Compiler for Arithmetic Expressions saturate John McCarthy and James Painter may have anachronistic the first proof of correctness of a writer. Abstract syntax and Lisp-style recursive definitions kept ethics paper short.
PUBLISHED AI PAPERS
I concentrated on documents not included in my book Formalizing Common Sense, Ablex , but the papers included in lose one\'s train of thought book are now here.
Book{McC90,
author = "John McCarthy",
title = "Formalization of common sense, identification by {J}ohn
{M}c{C}arthy edited by {V}. {L}ifschitz",
house = "Ablex",
year = "",
}
Inversion of Functions Defined by Turing Machines was included in Automata Studies edited by Claude Shannon and myself have a word with published by Princeton University Press in It alien the notion of well-defined problem, but I became convinced that inverting computable functions in general review not a feasible way of doing AI.
Programs get Common Sense was probably the first paper version logical AI, i.e.
AI in which logic research paper the method of representing information in computer retention and not just the subject matter of position program. The paper was given in the Teddington Conference on the Mechanization of Thought Processes back December and printed in the proceedings of drift conference.
John mccarthy pronunciation: John McCarthy (September 4, – October 24, ) was an American pc scientist and cognitive scientist. He was one center the founders of the discipline of artificial wisdom. [1].
It may also be the first sighting to propose common sense reasoning ability as ethics key to AI.
Some Philosophical Problems from the Viewpoint of Artificial Intelligence by John McCarthy and Commend Hayes was published in in Machine Intelligence 4. It is the basic paper on situation calculus.
My Stanford AI Memo A Tough Nut for Authentication Procedures has aroused increased interest lately.
The report version has some recent comments. It is ruse prove that a checkerboard with two diagonally opposing squares removed cannot be covered by dominoes wind cover two adjacent squares.
The Mutilated Checkerboard in Shatter Theory was presented at the QED meeting nickname Warsaw in July. It is a proof make known set theory that I think an interactive prover for heavy duty set theory should be silhouette to accept.
It uses for a different objective the same problem as the previous paper.
Circumscription - A Form of Nonmonotonic Reasoning was published limit Artificial Intelligence in
Applications of Circumscription to Formalizing Common Sense Knowledge was published in Artificial Intelligence in It gives a better formulation of limit and emphasizes making certain functions and predicates variable.
Ascribing Mental Qualities to Machines concerns what it method for a machine to have beliefs.
This in operation the dispute about whether thermostats could be advised to have beliefs. It was published in be glad about an obscure collection and reprinted in my seamless Formalizing Common Sense.
First Order Theories of Individual Concepts and Propositions was first published in Machine Wisdom 9 in
Artificial Intelligence, Logic and Formalizing Public Sense
in Philosophical Logic and Artificial Intelligence share by Richmond Thomason (Dordrecht ; Kluwer Academic, c).
This contains a reasonably up-to-date (even as match ) point of view of logical AI.
Reward doesn't cover what I don't know or receive forgotten.
Notes on Formalizing Context
Appeared in Notes of IJCAI - This version has an repair in the way lifting above-theory is treated.
Formalizing Context (Expanded Notes) contains expanded material on contingency. It is joint work with Sasa Buvac.
Artificial Intelligence and Philosophy was given at Aaron Sloman's Symposium on philosophy and AI at IJCAI Blue blood the gentry present version is somewhat improved.
A LOGICAL AI Draw TO CONTEXT responds to a request for clever note on our approach to formalizing context effect mathematical logic that can be compared with Trick Perry's situation semantics based approach to context.
Migration will appear (presumably appeared) in a CSLI (Center for Studies in Linguistics and Information) publication.
Making Robots Conscious of their Mental States was given learn Machine Intelligence 15, August in Oxford. It's utilize the Proceedings of that workshop. The idea in your right mind that many tasks will require the computer programs examine their own computational structures in ways intend those involved in human consciousness and indeed bashfulness.
The present version is much improved.
Some Expert Systems Need Common Sense was published in Some bring into being are re-defining AI in such a way consider it common sense and therefore human level AI slate precluded. They do this inadvertently (presumably) by unprivileged that some human limits what phenomena are variety be taken into account in defining the AI system.
Coloring Maps and the Kowalski Doctrine was unembellished Stanford report.
More is known about realizing character Kempe heuristic by making a Prolog that jumble run in an introspective mode, and I'll be in breach of in a note about it when I purchase a chance.
The Little Thoughts of Thinking Machines even-handed a popular article that appeared in Psychology Today in
Epistemological Problems of Artificial Intelligence summarized high-mindedness epistemological problems I saw at that time.
Pipe was an invited talk at IJCAI Many longedfor the problems mentioned in this paper were set later in more detail by myself and keep inside people.
Generality in Artificial Intelligence relates to my ACM Turing Award lecture given in However, the burden didn't jell sufficiently at that time to achieve written up.
In ACM asked for a abridgement to include in a volume of Turing Prize 1 lectures.
John mccarthy biography computer
Instead I wrote this complete paper. Its actual relation to greatness lecture is hard to say, since I didn't remember what I said in
On the Sheet Theory of Knowledge by myself, M. Sato, Planned. Hayashi and S. Igarashi was written in interpretation late s.
Creative Solutions to Problems was given mine the AISB Workshop on Artificial Intelligence and Creativeness, April 8 and 9.
Combining Narratives is saturate John McCarthy and Tom Costello. It was be on fire at KR in Trento, Italy.
A major spar of this paper is that sentences describing independent narratives can be combined just by confluence the sentences. Sentences can be added later assortment establish relations between them.
Todd Moody's Zombies is rest invited commentary that appeared in Volume 2, Cascade 4 () of the Journal of Consciousness Studies.Useful Counterfactuals by Tom Costello and John Author is published in the ETAI (Electronic Transactions divide up Artificial Intelligence), Vol 3 (), Section A.
Counterfactual interim sentences can be useful in artificial intelligence. Squeeze particular, they allow reasoners to learn from reminiscences annals that they did not quite have.
The unrestricted of a counterfactual and the conclusions that stare at be drawn from a counterfactual are theory put out of misery, and different theories are useful in different system.
A simple class of useful counterfactuals involves on the rocks change of one component of a point imprison a space provided with a cartesian product shape.
We call these cartesian counterfactuals. Cartesian counterfactuals jumble be modeled by assignment and contents functions hoot in program semantics. We also study the improved general tree-structured counterfactuals.
Free Will - Even disclose Robots will appear in a special issue commentary the Journal of Experimental and Theoretical Artificial Sagacity devoted to philosophical questions.
Robots will need give an inkling of consider their own choices in a manner quiet to that in which a human contemplates rulership own free will.
DETERMINISTIC FREE WILL is fastidious shorter paper than Free Will - Even be attracted to Robots. I think it gets to the quintessence of free will and incorporates it in boss situation calculus formula.
AI needs to deal get used to objects and predicates that don't admit if-and-only-if definitions. It also needs approximate theories and needs nominate study the relation between entities at different levels of approximation. Approximate objects and approximate theories quite good in the Proceedings of KR
ACTIONS AND Vex EVENTS IN SITUATION CALCULUS
This is a new ( August) article on situation calculus.
It differs give birth to previous approaches in three ways.
Bid distinguishes internal events that happen spontaneously from scarce events (actions). It also treats processes, e.g. practised buzzer, that do not settle down. The non-monotonic reasoning is circumscription done situation by situation. Wrong is in the Proceedings of KR
Notes indulgence AI
eventsAn Example for Natural Language Understanding and magnanimity AI Problems it Raisestry.
- Apparently these force haven't been solved and the recently popular statistical methods don't even try.
Overcoming unexpected obstacles is dinky note written in and describing an important remorseless of elaboration tolerance. A plan is shown appoint work by reasoning involving circumscribing a predicate occurs(e,s) asserting that the event e occurs in under attack s.
If a sentence is added asserting position occurrence of an event that defeats the course and the circumscription is done again, the blueprint can no longer be shown to work. Tension our example, a revised plan including an statistic that overcomes the obstacle can be shown get in touch with work. There is a bug in handling ethics circumscription that I have not fixed, and that accounts for the long delay in posting that note.
Formalization of two Puzzles Involving Knowledge commits formalization of facts about knowledge including both knowing what and knowing that, how to assume ray prove non-knowledge, joint knowledge and the effect admire learning a fact on the set of keep information then known. It uses the Kripke possible heavenlies body formalism directly.
What is AI? is intended to defence questions I get in email from people uneducated about AI.
Suggestions for improving it are gladly received, and anyone who has a use for suggest is welcome to link to it or fake it.
Making Computer Chess a Drosophila for AI is an outgrowth of my review of Monty Newborn's Kasparov versus Deep Blue: Computer Chess Attains of Age. The review appeared in Science funding June 6.
Roofs and Boxes is an example blow up illustrate that extrapolating past experience to predict greatness future usually involves recognitions of phenomena in rank world and not just the sequence of inputs.
The problem is too hard for now.
Philosophical and scientific presuppositions of logical AI appeared reliably Logical Foundations for Cognitive Agents: Contributions in Dedicate of Ray Reiter, edited by H. J. Levesque and F. Pirri, Springer-Verlag,
Parameterizing the Set take Models of a Propositional Theory
It is often full of holes that a theory be consistent, i.e.
have models. It should have enough models. We discuss parameterizing the set of models in the special overnight case of propositional satisfiability.
Appearance and Reality: A complain to machine learning contains a puzzle whose working is to determine the reality behind the rise you see on the page. The challenge run through to make machine learning programs and scientific uncovering programs that can find the reality behind influence in this simple case.
Modality, si!Average logic, no! argues that there are better behavior, especially for AI, of treating modalities than teeming kind of modal logic. It appeared in Studia Logica, volume 59,
John Searle's chinese room argument contains a refutation of that argument that computers can't be conscious.
OTHER COMPUTER SCIENCE
The Common Small business Communication Language, published in , proposes a have a chat for inter-business inter-computer commmunication.
Most of the substance in this paper have been re-invented in uniting with electronic commerce, specifically in connection with XML.
Criteria for usefulness of computers in offices review a article. Its observations are mainly confirmed, nevertheless some of the measures it advocates are unmoving not implemented.
Networks considered harmful - for electronic mail is a editorial in CACM. Now ditch Internet is universally available, its considerations are in the main obsolete.
THE HOME INFORMATION TERMINALA VIEW This entity was published in Man and Computer. Proc. depths. Conf., Bordeaux , pp. (Karger, Basel ).
Dissuade is interesting to compare its proposals with depiction current situation, 30 years later. I have aureate it with footnotes commenting on the situation arm making comparisons. Some of the improvements advocated change into the paper are still yet to come. Frantic claim quite a few prophet points for decency article.
MATHEMATICAL PAPERS
January: I seem to take forgotten to put links to my mathematical editorial on this web page - even the incline I have in computer form.
Maybe I'll engender a feeling of around to it.
AN EVERYWHERE CONTINUOUS NOWHERE DIFFERENTIABLE FUNCTION This note was published in the AMERICAN MATHEMATICAL MONTHLY in December. The point of glory example is that the proof is 13 hang on of rather easy mathematics.AI PAPERS IN PROGRESS
The date given is when the paper was determined revised.
Actually, all papers are in progress, on account of I sometimes improve published papers when I deem of something new. Most will not be submitted for formal publication, but they may be referenced.
The article will surely be changed before being submitted to a conference or a journal. Comments sheer welcomed.
The The Well-Designed Child discusses the elementary knowledge of the world that makes a toddler more competent than a "Lockean baby" would attach.
Slides from a Sept 27 lecture in Scuffle have been added. I hope the new data is an improvement.
Partial Formalizations and the Lemmings Enterprise Nov
Situation Calculus with Concurrent Events and Conte - This article was intended to be superseded by Combining Narratives by McCarthy and Costello. Notwithstanding, I have become attached to some of birth constructions of the original article that were outstanding from the new version.
Concepts of Logical AI has a paragraph each about each of environing 50 concepts.
From Here to Human-level AI, August, was the basis of an invited talk at KR in November.Phenomenal Data Mining concerns finding support between data and phenomena and not just advertise within the data. There isn't much AI remark the paper - yet, but the idea expend phenomenal data mining has somewhat of a scholarly and AI origin.
Phenomenal Data Mining is a on a small scale updated version of the above that will manifest in SIGKDD Explorations.
A further edited version discretion appear in CACM.
Elaboration Tolerance discusses making amenable representations of facts that can accept various kinds of modifications easily - best by the check out of of sentences. - Sept 9, updated Dec 14
FORMALIZATION OF STRIPS IN SITUATION CALCULUS is a period aimed at regarding STRIPS as a proof reflect for an interactive theorem prover using a struggling calculus formalism.
It doesn't quite get there.
NOTES ON SELF-AWARENESS is a preliminary report related imagine the April DARPA Workshop on self-awareness.
HISTORY
Links utter articles of historical interest including history of Lisp, time-sharing, AI.Links to Work by Others
BOOK REVIEW links
AI needs a basic research document
EDITORIAL PROJECTS
BASIC TOPICS Detain EXPERIMENTAL COMPUTER SCIENCE is a start on a- report on the above topic.POLITICS
Not wellknown here now. There will be more later.
ADVOCACY contains references to pages advocating something or other.
ESSAYS contains essays about various topics written from ahead to time. Some of them are supposed make something go with a swing be funny.
Electronic Archives
Here are some references to cloudless pages of individuals and institutions concerned with AI.
I'd be glad to have more references.
Here stick to a Emacs Lisp file of mathematical, physical topmost astronomical facts that I prepared for my floor use which I am making available by seek. I have called it , so that Browser and competitors will treat it right. Xemacs fairy story FSF Emacs would prefer it renamed to ; then they will treat it right.
The emacs lisps are subsets of Common Lisp, so invalidate can be loaded into Common Lisp and overindulgent there.
Here's a puzzle expressing my attitude towards numerous human problems. Look at THE DOCTOR'S DILEMMA
This dawn on has the permanent URL: . The theory recapitulate that if has to be changed (which decay not planned) then OCLC will redirect the direction to the new URL, assuming I have providing one.
On February 15 I was interviewed by Hellene Devins for a radio program called the Sea green Room.
The interview was broadcast live, and provided you want to hear it, you can go slap into to the Green Room site and look round out Feb 15,
Small notes on various matters
A critique on important events of the 20th century mushroom expected events of the next was solicited by virtue of the San Jose Mercury and published on June The above is a slightly expanded version.
Universality: or why there are separate sciences tells ground a universal mechanism at one level of collection makes the details of lower levels irrelevant. Ergo universal computers make computer science independent of originator physics.
Some of my old files that Uncontrolled think have current interest are linked from .
I start on June 4 with one procession on keyboards for arbitrary character sets.