Aritifical_Intelligence_Chapter1_INTRODUCTION

Aritifical_Intelligence_Chapter1_INTRODUCTION

1. Chapter 1 Introduction 2 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited

2. Artificial Intelligence We humans not only want to understand minds, brains and intelligence, but also want to create minds, brains and intelligence. This quest goes broadly under the name of artificial intelligence. We call programs 'intelligent', if they exhibit behaviours that would be regarded intelligent if they were exhibited by human beings—Herbert Simon. AI is the study of mental faculties through the use of computational models—Eugene Charniak and Drew McDermott. The fundamental goal of this research is not merely to mimic intelligence or produce some clever fake. "AI" wants the genuine article; machines with minds—John Haugeland (1985). 3 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited

3. Historical Backdrop The urge to create intelligent beings can be traced back to Greek mythology. Hephaestus, son of Hera (the queen of gods and Zeus' wife), constructed humanlike creations regularly in his forge. Heron of Alexandria in the first century AD built water powered mechanical ducks that emitted realistic chirping sounds. Pope Sylvester II (946¬1003) is said to have made a statue with a talking head, with a limited vocabulary, and a penchant for predicting the future. It gave replies to queries with a yes or a no, and its human audience did not doubt that the answer was preceded by some impressive mental activity.

4. The word robot, which means worker in Czech, first appeared in a play called Rossum's Universal Robots (RUR) by Karel Capek (1921) Isaac Asimov formulated the three laws of robotics : A robot cannot injure a human being, or through inaction allow a human being to come to harm. A robot must obey any orders given to it by human beings, except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law. A zeroeth law was subsequently added: A robot may not injure humanity, or, by inaction, allow humanity to come to harm. 5 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited Historical Backdrop (contd.)

5. 6 Initially, the world we saw was made up of a flat earth, with the heavens rotating around it. See figure Galileo Galilei said when we taste, smell or hear something and associate a name with it, the name is really a name we give to the sensation that arises in us. Hobbes had given up on the idea of thoughts being in the image of reality. That is a question that we can say is still unresolved. Hobbes parcels were material in nature. 6 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited Mind and Body

6. 7 The first checkers playing program being written in 1951 in the University of Manchester by Christopher Strachey (1916-1975). Samuel wrote first program that incorporated learning. Samuel also invented Alpha-Beta pruning, a method that can drastically cut down on the search effort. Alex Bernstein had developed the first chess playing program in 1957. A natural language parser developed in 1963 by Susumu Kuno at Harvard revealed the degree of ambiguity in English that often escapes human listeners, who latch on to one parse tree and the corresponding meaning. In 1966, the US government appointed Automatic Language Processing Advisory Committee (ALPAC). In 1961, James Slage wrote the first symbolic integration program, SAINT, which formed the base for many symbolic mathematics tools. 7 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited AI in the Last Century

7. 8 Meanwhile, John McCarthy, focused on logic in computer science, and proposed a system called Advice Taker in 1958, which was to use logic as a vehicle for knowledge representation and commonsense reasoning. He also invented Lisp . In 1965, Alan Robinson published the Resolution Method for theorem proving that brought all kinds of logical inferences into one uniform fold. This approach gave a tremendous fillip to logical theorem-proving research. The first autonomous robot, The Hopkin’s Beast was built at the Johns Hopkins University in the early sixties. The first planning system STRIPS (Stanford Research Institute Planning System) was developed around the same time, 1971, by Richard Fikes and Nils Nilsson. MYCIN, a medical diagnosis system, the doctoral work of Edward Shortliffe , appeared in 70s. 8 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited AI in the Last Century (contd.)

8. The ability to perceive the world around us, to procure food for satisfying one’s hunger, to pick a spot of shade to avoid the sun in the tropics, or a spot of warm sun in a cooler climate, to sense danger and flee, or fight if escape is not possible, are surely characteristics of intelligent behaviour . The computer is only a symbol manipulator. It does not know what it is doing. It does not have a sense of being in this world. The philosophical objection to artificial intelligence is that since computers are not capable of self awareness, being just programmed machines, they cannot be called intelligent. One would have observed that definitions of intelligence are usually human centric. This may be simply because as humans, we can see life only from our own perspective. 9 9 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited What is Intelligence?

9. Perhaps, in order to circumvent the debate on intelligence and get on with the business of building machines that are intelligent, Alan Turing, prescribed a test to determine whether a machine (computer program) is intelligent (Turing, 1950). A chatterbot program called ELIZA , written by Joseph Weizenbaum (1966) in 1966, showed how human judgment is fallible. The most popular version of ELIZA is a program running the script Doctor , and can be found amongst other places in the emacs editor. One well known objection to the Turing Test was put forward by John Searle (1980) who says that programs like ELIZA could pass the Turing Test without understanding the words they were using at all. One must keep in mind that ELIZA itself has no understanding of words, but it cannot also be said to have passed the test. 10 10 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited The Turing Test

10. An agent can do the right thing only when the agent knows the situation. Intelligence is perhaps most associated with the faculty of understanding; the ability to perceive the world around us and comprehend the meanings of signs and symbols, including language; the ability to grasp the situation. Actions or decisions can only be judged in the context of the goal. Intelligence (or wisdom, in the larger context) may often lie in the selection of the goal to pursue. Intelligence, thus, may manifest itself in the choice of goals an agent may choose, and the means adopted by the agent to achieve those goals. It also involves the ability to understand the situation and reason about the consequences of one’s actions. It involves learning from experiences & adapting one’s behaviour in a dynamic environment. 11 11 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited Intelligent Decisions

11. The distinctive features of intelligence can perhaps be summed up in one word—imagination. Figure shows : An intelligent agent in a world carries a model of the world in its “head”. The model may be an abstraction. A self aware agent would model itself in the world ‘model’. Deeper awareness may require that the agent represents (be aware of) itself modeling the world. In Advaita Vedanta , ‘Maya’ is used to refer to the limited mental and physical reality that our consciousness has got entangled in. Maya creates the illusion of the physical universe and lead us to perceive a rope to be a snake. 12 12 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited The Bottom Line

12. Topics in AI 13 13 A sampling of the “society of the mind”. Different kinds of reasoning methods would be required by an intelligent mind, as it goes through a sense-deliberate-act cycle. The dashed circle represents symbolic reasoning in artificial intelligence. Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited

13. 14 14 A model of a cognitive agent. The innermost circle represents Classical Al that is concerned with symbolic reasoning. This is encapsulated in neuro-fuzzy systems that produce and consume the symbols. The outermost layer deals with the external world directly, processing signals and producing motor activity. Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited Topics in AI (contd.)

14. Incidentally, one must realize that the distinction between analog and digital is entirely of our own making, and mostly reflects the nature of our devices. The earliest radios were made up of vacuum tubes, as were the earliest computers until the invention of the diode. Anything that we do on a digital computer has to be symbolic at the deepest level, because the digital machine can in the end distinguish between only two kinds of symbols—0 and 1. One can imagine that the core symbolic reasoning system has a shell of processes around it that converts signals to symbols and vice versa. We will not deal with neural networks and similar systems in which we cannot interpret weights of edges, and what a node represents. We will be concerned, however, with networks of nodes, semantic nets, where each node will represent something meaningful. 15 15 Chapter 1 Copyright © 2013 by McGraw Hill Education (India) Private Limited Topics in AI (contd.)