AIBook.01 | Intelligence | Information

Lecture Overview

What is Artificial Intelligence? Agents acting in an environment

c D. Poole and A. Mackworth 2010

Artificial Intelligence, Lecture 1.1, Page 1

Page 2 . Poole and A. Mackworth 2010 Artificial Intelligence.What is Artificial Intelligence? Artificial Intelligence is the synthesis and analysis of computational agents that act intelligently.1. An agent acts intelligently if: its actions are appropriate for its goals and circumstances it is flexible to changing environments and goals it learns from experience it makes appropriate choices given perceptual and computational limitations c D. An agent is something that acts in an environment. Lecture 1.

Poole and A. build. Mackworth 2010 Artificial Intelligence. Lecture 1. intelligent artifacts. Page 3 . Analogy between studying flying machines and thinking machines. c D.Goals of Artificial Intelligence Scientific goal: to understand the principles that make intelligent behavior possible in natural or artificial systems. and experiment with computational systems that perform tasks that require intelligence Engineering goal: design useful. analyze natural and artificial agents formulate and test hypotheses about what it takes to construct intelligent agents design.1.

Lecture 1. Mackworth 2010 Artificial Intelligence. Poole and A.Agents acting in an environment Abilities Goals/Preferences Prior Knowledge Observations Actions Past Experiences Environment Agent c D. Page 4 .1.

. score goals. speech. speech recognition. Mackworth 2010 Artificial Intelligence. explore. observations: vision. what a sensor tell us. . sound. ... . categories of objects.Example agent: robot abilities: movement. Lecture 1. gesture recognition. grippers. Poole and A. past experiences: effect of steering. ... . . . slipperiness. how people move. . c D. rescue people.. Page 5 . prior knowledge: what is important feature. sonar. goals: deliver food.1. facial expressions. .

. .Example agent: teacher abilities: present new concept.. social skills. . skills. goals: particular knowledge. focus. drill. . explain concept. . . teaching strategies. observations: test results. inquisitiveness. c D.. Lecture 1.. Mackworth 2010 Artificial Intelligence. give test. effects of teaching strategies. . . prior knowledge: subject material. errors.1.. facial expressions. . Page 6 . Poole and A. past experiences: prior test results. . .

. . increase life expectancy. past experiences: treatment outcomes. . prior knowledge: possible diseases. effects of drugs. test results. possible causal relationships. goals: remove disease. observations: verbal symptoms. . . .1. Lecture 1. Poole and A. . . . Page 7 .. prescribe drugs. test results given symptoms. relieve pain. test. explain instructions. c D. reduce costs. .. symptoms. visual appearance.Example agent: medical doctor abilities: operate. Mackworth 2010 Artificial Intelligence.

... . information retrieved. reliability of information sources. facial expressions. Mackworth 2010 Artificial Intelligence. Lecture 1.. . presentation modalities. . . goals: present information. filter information. privacy. interrupt. . . . observations: users request. maximize useful information. user feedback. past experiences: effect of presentation modes. Page 8 . Poole and A. minimize irrelevant information. find another information source. . ask user.Example agent: user interface abilities: present information. c D. .1. prior knowledge: information sources.

Diagnostic assistant helps a human troubleshoot problems and suggests repairs or treatments. Mackworth 2010 Artificial Intelligence.3. E. Trading agent buys goods and services on your behalf. Poole and A. Intelligent tutoring system teaches students in some subject area. medical diagnosis. .Four Example Application Domains Autonomous delivery robot roams around an office environment and delivers coffee.g. electrical problems.. Page 1 . Lecture 1. parcels. c D. ..

Poole and A. Lecture 1. Mackworth 2010 Artificial Intelligence.3.Domain for Delivery Robot r131 r129 r127 r125 r123 r121 r119 r117 lab4 lab3 r115 lab1 lab2 r113 stairs r101 r103 r105 r107 r109 r111 c D. Page 2 .

Mackworth 2010 Artificial Intelligence. Poole and A. laser range finders. pickup and place objects. tradeoffs between acting quickly and acting safely.3. Goals: what it needs to deliver and when. what objects are there. Observations: about its environment from cameras. maps. Lecture 1. sonar. objects it may encounter. Past experience: which actions are useful and when.Autonomous Delivery Robot Example inputs: Abilities: movement. how its actions affect its position. speech. sound. or keyboards. Page 3 . Prior knowledge: its capabilities. c D.

Page 4 . Plan how to carry out multiple tasks.What does the Delivery Robot need to do? Determine where Craig’s office is.3. . Make tradeoffs under uncertainty: should it go near the stairs? Learn from experience. c D. . Mackworth 2010 Artificial Intelligence. Where coffee is. avoid obstacles. Lecture 1. Sense the world. Make default assumptions about where Craig is. pickup and put down coffee. Find a path between locations. Poole and A.

Domain for Diagnostic Assistant cb1 s1 s2 w0 w4 l1 l2 p1 w1 w2 s3 w5 cb2 off w6 p2 on circuit breaker outside power w3 switch two-way switch light power outlet c D. Poole and A. Mackworth 2010 Artificial Intelligence. Lecture 1.3. Page 5 .

Past experience: the effects of repairs or treatments. Lecture 1.3. c D.Diagnostic Assistant Example inputs: Abilities: recommends fixes. Observations: symptoms of a device or patient. ask questions. Poole and A. Page 6 . the side effects of repairs. how malfunctions manifest themselves. what information tests provide. the prevalence of faults or diseases. Mackworth 2010 Artificial Intelligence. Goals: fixing the device and tradeoffs between fixing or replacing different components. Prior knowledge: how switches and lights work.

Page 7 . Lecture 1.3. Search through the space of possible fault complexes. Mackworth 2010 Artificial Intelligence. Poole and A. the effects of treatments. and the accuracy of tests. Plan courses of tests and treatments to address the problems. Reason about the uncertainties/ambiguities given symptoms. Learn what symptoms are associated with faults. c D. Derive possible causes for symptoms. Explain its reasoning to the human who is using it. rule out other causes.Subtasks for the diagnostic assistant Derive the effects of faults and interventions. Trade off alternate courses of action.

Page 8 . Lecture 1. and some items may be useless without other items. Mackworth 2010 Artificial Intelligence. c D.3. It finds what is available. desires and preferences. It purchases goods and services that fit together to fulfill your preferences. Poole and A.Trading Agent Trading agent interacts with an information environment to purchase goods and services. It acquires a users needs. It is difficult because users preferences and what is available can change dynamically.

3. who has good deals. Mackworth 2010 Artificial Intelligence. where to purchase items. Goals: what the person wants. purchase items. how to decompose a complex item. Poole and A. Past experience: how long special last. c D. Prior knowledge: the ontology of what things are available. number in stock. prices. how long items take to sell out. Lecture 1. their tradeoff. Observations: what items are available. make recommendations.Trading Agent Inputs Abilities: acquire information. Page 9 . what your competitors do.

Poole and A. interest Observations: test results. Page 10 . what the student is concentrating on c D. Lecture 1. study skills.3. gain social skills. Mackworth 2010 Artificial Intelligence. primitive strategies Past experience: common errors. inquisitiveness. questions.Intelligent Tutoring System Abilities: Present information. effects of teaching strategies Goals: the students should master subject material. facial expressions. give tests Prior knowledge: subject material.

determine what the world is like.Common Tasks of the Domains Modeling the environment Build models of the physical environment. c D. Lecture 1. patient. Mackworth 2010 Artificial Intelligence. Page 11 . Action Given a model of the world and a goal.3. Learning from past experiences Learn about the specific case and the population of cases. determine what should be done. Poole and A. Evidential reasoning or perception Given observations. or information environment.

Sign up to vote on this title
UsefulNot useful