Course Assistants CS 121: Introduction to AI Jean-Claude Latombe � Jacob Quain ai.stanford.edu/~latombe � Nikil Viswanathan cs121.stanford.edu Required textbook: S. Russell and P. Norvig. Artificial Intelligence: A Modern Approach. 3 rd edition, Prentice Hall, 2010 1 2 Office Hours and Sections Today’s Agenda JCL Mon at 11am-12pm in Gates 135 � � Introduction to AI Jacob Quain � (Russell and Norvig: Chap. 1 and 2) Nikil Viswanathan Nikil Viswanathan � � Overview of CS121 O i f CS121 CA section: � 3 4 What is AI? What is AI? (R&N) an attempt of � AI is the reproduction of human Discipline that systematizes and automates reasoning and intelligent behavior by reasoning processes to create machines that: computational methods Act like humans A t lik h Act rationally A t ti ll Intelligent behavior Think like humans Think rationally Computer Humans 5 6 1
Act like humans Act rationally Act like humans Act rationally Think like humans Think rationally Think like humans Think rationally � The goal of AI is to create computer systems that perform tasks regarded as requiring intelligence when done by humans � Here, how the computer performs tasks does matter � � AI Methodology: Take a task at which people are b better, e.g.: � The reasoning steps are important • Prove a theorem • Play chess � � Ability to create and manipulate symbolic • Plan a surgical operation knowledge (definitions, concepts, theorems, …) • Diagnose a disease • Navigate in a building � What is the impact of hardware on low-level and build a computer system that does it automatically reasoning, e.g., to go from signals to symbols? � But do we want to duplicate human imperfections? 7 8 Act like humans Act rationally Can Machines Act/Think Think like humans Think rationally Intelligently? “If there were machines which bore a resemblance to � Now, the goal is to build agents that always make the our bodies and imitated our actions as closely as “best” decision given what is available (knowledge, possible for all practical purposes, we should still have time, resources) two very certain means of recognizing that they were not real men. The first is that they could never use � “Best” means maximizing the expected value of a words, or put together signs, as we do in order to utility function declare our thoughts to others… Secondly, even though � � Connections to economics and control theory some machines might do some things as well as we do � What is the impact of self-consciousness, emotions, them, or perhaps even better, they would inevitably desires, love for music, fear of dying, etc ... on human fail in others, which would reveal that they are acting intelligence? not from understanding, …” Discourse on the Method, by Descartes (1598-1650) 9 10 Can Machines Act/Think An Application of the Turing Test Intelligently? � CAPTCHA: Completely Automatic Public Turing Test: Turing tests to tell Computers and http://plato.stanford.edu/entries/turing-test/ � Humans Apart Test proposed by Alan Turing in 1950 � � E g : � E.g.: The computer is asked questions by a human The computer is asked questions by a human � � interrogator. It passes the test if the interrogator cannot tell whether the responses • Display visually distorted words come from a person • Ask user to recognize these words Required capabilities: natural language � processing, knowledge representation, � Example of application: have only automated reasoning, learning,... humans open email accounts No physical interaction � Chinese Room (J. Searle) � 11 12 2
Can Machines Act/Think Some Achievements Intelligently? Computers have won over world � champions in several games, including Checkers, Othello, and Chess, but still � Yes, if intelligence is narrowly defined as do not do well in Go AI techniques are used in many � information processing systems: formal calculus, video games, route planning, logistics planning, pharmaceutical drug design, medical diagnosis, hardware and software AI has made impressive achievements showing that trouble-shooting, speech recognition, traffic monitoring, recognition traffic monitoring tasks initially assumed to require intelligence can be k ll d ll b facial recognition, automated medical image analysis, part inspection, etc... Stanford’s robotic car, Stanley, But each success of AI seems to push further the limits � autonomously traversed 132 miles of what we consider “intelligence” of desert Some industries (automobile, � electronics) are highly robotized, while other robots perform brain and heart surgery, are rolling on Mars, fly autonomously, …, but home robots still remain a thing of the future 13 14 Can Machines Act/Think Some Big Open Questions Intelligently? � AI (especially, the “rational agent” approach) assumes � Yes, if intelligence is narrowly defined as that intelligent behaviors are only based on information processing? Is this a valid assumption? information processing � If yes, can the human brain machinery solve problems that are inherently intractable for computers? AI has made impressive achievements showing that tasks initially assumed to require intelligence can be k ll d ll b � In a human being, where is the interface between automated “intelligence” and the rest of “human nature”, e.g.: • How does intelligence relate to emotions felt? � Maybe yes, maybe not, if intelligence is • What does it mean for a human to “feel” that he/she not separated from the rest of “being understands something? � Is this interface critical to intelligence? Can there human” exist a general theory of intelligence independent of human beings? What is the role of the human body? 15 16 Some Big Open Questions � AI contributes to building an information processing model of human beings, just as Biochemistry contributes to building a model � AI (especially, the “rational agent” approach) assumes of human beings based on bio-molecular that intelligent behaviors are based on information In the movie I, Robot, the most impressive processing? Is this a valid assumption? interactions feature of the robots is not their ability to � If yes, can the human brain machinery solve problems � Both try to explain how a human being that are inherently intractable for computers? solve complex problems, but how they blend op rat s operates human-like reasoning with other key � In a human being, where is the interface between � Both also explore ways to avoid human “intelligence” and the rest of “human nature”, e.g.: aspects of human beings (especially, self- imperfections (in Biochemistry, by engineering new � How does intelligence relate to emotions felt? consciousness, fear of dying, distinction proteins and drug molecules; in AI, by designing � What does it mean for a human to “feel” that he/she between right and wrong) rational reasoning methods) understands something? � Both try to produce new useful technologies � Is this interface critical to intelligence? Can there exist a general theory of intelligence independent of � Neither explains (yet?) the true meaning of human beings? What is the role of the human body? being human 17 18 3
Recommend
More recommend
