LCC 3710Principles of Interaction DesignClass agenda:-Readings- Group Activity- AI, Agents, Artificial LifeReadingsTuring, Alan (1950). "Computing Machinery and Intelligence" inMind, A Quarterly Review of Psychology and Philosophy, Vol. LIX,No. 236, October 1950.Maes, Patti (1995). "Artificial Life Meets Entertainment: LifelikeAutonomous Agents" in Communications of the ACM, Vol. 38, no.11, November 1995.Moggridge, Will WrightWhat is AI?Big QuestionsCan machines think?And if so, how?And if not, why not?And what does this say about human beings?And what does this say about the mind?What is AI?There are no crisp definitionsHere’s one from John McCarthy, (He coined the phrase AI in 1956) -see http://www.formal.Stanford.EDU/jmc/whatisai/)Q. What is artificial intelligence?A. It is the science and engineering of making intelligent machines,especially intelligent computer programs. It is related to the similartask of using computers to understand human intelligence, but AIdoes not have to confine itself to methods that are biologicallyobservable.Q. Yes, but what is intelligence?A. Intelligence is the computational part of the ability to achieve goalsin the world. Varying kinds and degrees of intelligence occur inpeople, many animals and some machines.Other DefinitionsAI is a collection of hard problems which can be solved byhumans and other living things, but for which we don’t havegood algorithmic solutionsE.g. understanding spoken natural language, medical diagnosis,circuit design, etc.AI Problem + Sound theory = Engineering problemSome problems used to be thought of as AI but are nowconsidered not AIE.g. compiling Fortran in 1955, symbolic mathematics in 1965What’s easy and what’s hard?It’s been easier to mechanize many of the high level tasks weusually associate with "intelligence" in peopleE.g. Symbolic integration, proving theorems, playing chess,medical diagnosis etc.It's been very hard to mechanize tasks that lots of animals can doWalking around without running into thingsCatching prey and avoiding predatorsInterpreting complex sensory information (e.g., visual, aural, …)Modeling the internal states of other animals from their behaviorWorking as a team (e.g. with pack animals)Is there a fundamental difference between the two categories?Foundations of AIComputerScience & EngineeringAIMathematicsCognitiveSciencePhilosophyPsychology LinguisticsBiologyEconomicsWhy AI?Engineering:To get machines to do a wider variety of useful thingsE.g. understand spoken natural language, recognize individual peoplein visual scenes, find the best travel plan for a vacation etc.Cognitive Science:As a way to understand how natural minds and mentalphenomena workE.g. visual perception, memory, learning, language etc.Philosophy:As a way to explore some basic and interesting (and important)philosophical questionsE.g. the mind body problem, what is consciousness etc.Possible ApproachesThinkActLike humans WellGeneralProblemSolverElizaRationalagentsHeuristicsystemsCurrent AItends to workmostly in thisareaDifferent approaches due to different criteria, two dimensions:Thought processes/reasoning vs. behavior/actionSuccess according to human standards vs. success according to anideal concept of intelligence: rationalityThinkActLike humans WellGPSElizaRationalagentsHeuristicsystemsThink wellDevelop formal models of knowledge representation,reasoning, learning, memory, problem solving, that can berendered in algorithms.There is often an emphasis on systems that are probablycorrect, and guarantee finding an optimal solution.Act wellFor a given set of inputs, generate an appropriate output that isnot necessarily correct but gets the job doneA heuristic (heuristic rule, heuristic method) is a rule of thumb,strategy, trick, simplification, or any other kind of device whichdrastically limits search for solutions in large problem spacesHeuristics do not guarantee optimal solutions or any solution atall: a useful heuristic merely offers solutions which are goodenough most of the timeThinkActLike humans WellGPSElizaRationalagentsHeuristicsystemsThink like humansCognitive science approach, focus not just on behavior and I/Obut also look at the reasoning process. Computational modelshould reflect "how" results were obtained.Provide a new language for expressing cognitive theories andnew mechanisms for evaluating themGPS (General Problem Solver): Goal is not just to producehumanlike behavior (like ELIZA), but to produce a sequence ofsteps of the reasoning process that are similar to the stepsfollowed by a person in solving the same task.ThinkActLike humans WellGPSElizaRationalagentsHeuristicsystemsAct like humansBehaviorist approach.Not interested in how you get results, just the similarity to whathuman results are.Exemplified by the Turing Test (Alan Turing, 1950).ThinkActLike humans WellGPSElizaRationalagentsHeuristicsystems1950 Turing TestAlan Turing (1950)"Can machines think?" ! "Can machines behave intelligently?"Operational test for intelligent behavior: the Imitation GameSeparate rooms contain a person, a computer, and aninterrogator. The interrogator can communicate with the othertwo by teleprinter. The interrogator tries to determine which isthe person and which is the machine.The machine tries to fool the interrogator into believing that it isthe person. If the machine succeeds, then we conclude that themachine can think.Turing TestPredicted that by 2000 a machine might have a 70%chance of fooling a lay person for 5 minutesAnticipated the major arguments against AI infollowing 50 yearsSuggested major components of AI:Knowledge, reasoning, language understanding,learningELIZAELIZA: A program that simulated a psychotherapistinteracting with a patient and successfully passed theTuring TestCoded at MIT during 1964-1966 by Joseph WeizenbaumFirst script was doctor:The script was a simple collection of syntactic patterns notunlike regular expressionsEach pattern had an associated reply which might include bitsof the input after simple transformations (my ! your)Weizenbaum was shocked at the reactions:Psychiatrists thought it had potentialPeople anthropomorphizedMany thought it solved the natural language problemIntelligent AgentsDefinitionAn agent perceives its environment via sensors andacts in that environment with its actuators oreffectors to maximize progress towards its goalsHence, an agent gets percepts one at a time, andmaps this percept sequence to actions (one action ata