SOLUTIONS & INSTRUCTOR MANUAL
Artificial Intelligence: A Modern Approach, 4th Edition
by Peter Norvig and Stuart Russell, Chapters 1 – 28
,Artificial Intelligence gt
gtgtgtgt 1 Introduction ...
gt gt gt
gtgtgtgt 2 Intelligent Agents ...
gt gt gt gt
II Problem-solving
gt
gtgtgtgt 3 Solving Problems by Searching ...
gt gt gt gt gt gt
gtgtgtgt 4 Search in Complex Environments ...
gt gt gt gt gt gt
gtgtgtgt 5 Adversarial Search and Games ...
gt gt gt gt gt gt
gtgtgtgt 6 Constraint Satisfaction Problems ...
gt gt gt gt gt
III Knowledge, reasoning, and planning
gt gt gt gt
gtgtgtgt 7 Logical Agents ...
gt gt gt gt
gtgtgtgt 8 First-Order Logic ...
gt gt gt gt
gtgtgtgt 9 Inference in First-Order Logic ...
gt gt gt gt gt
gtgtgtgt 10 Knowledge Representation ...
gt gt gt gt
gtgtgtgt 11 Automated Planning ...
gt gt gt gt
IV Uncertain knowledge and reasoning
gt gt gt gt
gtgtgtgt 12 Quantifying Uncertainty ...
gt gt gt gt
gtgtgtgt 13 Probabilistic Reasoning ...
gt gt gt gt
gtgtgtgt 14 Probabilistic Reasoning over Time ...
gt gt gt gt gt gt
gtgtgtgt 15 Probabilistic Programming ...
gt gt gt gt
gtgtgtgt 16 Making Simple Decisions ...
gt gt gt gt gt
gtgtgtgt 17 Making Complex Decisions ...
gt gt gt gt gt
gtgtgtgt 18 Multiagent Decision Making ...
gt gt gt gt gt
V Machine Learning
gt gt
, 19 Learning from Examples ...
gtgtgtgt gt gt gt gt gt
20 Learning Probabilistic Models ...
gtgtgtgt gt gt gt gt gt
21 Deep Learning ...
gtgtgtgt gt gt gt gt
22 Reinforcement Learning ...
gtgtgtgt gt gt gt gt
VI Communicating, perceiving, and acting
gt gt gt gt
23 Natural Language Processing ...
gtgtgtgt gt gt gt gt gt
24 Deep Learning for Natural Language Processing ...
gtgtgtgt gt gt gt gt gt gt gt gt
25 Computer Vision ...
gtgtgtgt gt gt gt gt
26 Robotics ...
gtgtgtgt gt gt gt
VII Conclusions
gt
27 Philosophy, Ethics, and Safety of AI ...
gtgtgtgt gt gt gt gt gt gt gt gt
28 The Future of AI
gtgtgtgt gt gt gt gt
, EXERCISES g t gt
1
INTRODUCTION
Notegtthatgtforgtmanygtofgtthegtquestionsgtingtthisgtchapter,gtwegtgivegtreferencesgtwheregtanswersgtc
angtbegtfoundgtrathergtthangtwritinggtthemgtout—thegtfullgtanswersgtwouldgtbegtfargttoogtlong.
1.1 What Is AI?
g t gt gt
Exercisegt1.1.#DEFA
Definegtingtyourgtowngtwords:g t (a)gtintelligence,gt(b)gtartificialgtintelligence,gt(c)gtagent,gt(d)gtr
a-gttionality,gt(e)gtlogicalgtreasoning.
a. Dictionarygtdefinitionsgtofgtintelligencegttalkgtaboutgt“thegtcapacitygttogtacquiregtandgtapp
lygtknowledge”gtorgt“thegtfacultygtofgtthoughtgtandgtreason”gtorgt“thegtabilitygttogtcompreh
endgtandgtprofitgtfromgtexperience.”g t Thesegtaregtallgtreasonablegtanswers,gtbutgtifgtwegtw
antgtsomethinggtquantifiablegtwegtwouldgtusegtsomethinggtlikegt“thegtabilitygttogtactgtsucces
sfullygtacrossgtagtwidegtrangegtofgtobjectivesgtingtcomplexgtenvironments.”
b. Wegtdefinegtartificialgtintelligencegtasgtthegtstudygtandgtconstructiongtofgtagentgtprogram
sgtthatgtperformgtwellgtingtagtgivengtclassgtofgtenvironments,gtforgtagtgivengtagentgtarchitect
ure;gttheygtdogtthegtrightgtthing.g t Angtimportantgtpartgtofgtthatgtisgtdealinggtwithgtthegtuncer
taintygtofgtwhatgtthegtcurrentgtstategtis,gtwhatgtthegtoutcomegtofgtpossiblegtactionsgtmightgtb
e,gtandgtwhatgtisgtitgtthatgtwegtreallygtdesire.
c. Wegtdefinegtangtagentgtasgtangtentitygtthatgttakesgtactiongtingtresponsegttogtperceptsgtfromgtan
gtenvi-gtronment.
d. Wegtdefinegtrationalitygtasgtthegtpropertygtofgtagtsystemgtwhichgtdoesgtthegt“rightgtthing”gt
givengtwhatgtitgtknows.g t SeegtSectiongt2.2gtforgtagtmoregtcompletegtdiscussion.g t Thegtbasi
cgtconceptgtisgtperfectgtrationality;gtSectiongt??gtdescribesgtthegtimpossibilitygtofgtachieving
gtperfectgtrational-gtitygtandgtproposesgtangtalternativegtdefinition.
e. Wegtdefinegtlogicalgtreasoninggtasgtthegtagtprocessgtofgtderivinggtnewgtsentencesgtfromgtold,gt
suchgtthatgtthegtnewgtsentencesgtaregtnecessarilygttruegtifgtthegtoldgtonesgtaregttrue.gt(Noticegtth
atgtdoesgtnotgtrefergttogtanygtspecificgtsyntaxgtorgtformalgtlanguage,gtbutgtitgtdoesgtrequiregtagtwel
l-definedgtnotiongtofgttruth.)
Exercisegt1.1.#TURI
ReadgtTuring’sgtoriginalgtpapergtongtAIgt(Turing,gt1950).g t Ingtthegtpaper,gthegtdiscussesgtsevera
lgtobjectionsgttogthisgtproposedgtenterprisegtandgthisgttestgtforgtintelligence.gtWhichgtobjectionsgtstillgtca
rry
©gt2023gtPearsongtEducation,gtHoboken,gtNJ.gtAllgtrightsgtres
erved.
Artificial Intelligence: A Modern Approach, 4th Edition
by Peter Norvig and Stuart Russell, Chapters 1 – 28
,Artificial Intelligence gt
gtgtgtgt 1 Introduction ...
gt gt gt
gtgtgtgt 2 Intelligent Agents ...
gt gt gt gt
II Problem-solving
gt
gtgtgtgt 3 Solving Problems by Searching ...
gt gt gt gt gt gt
gtgtgtgt 4 Search in Complex Environments ...
gt gt gt gt gt gt
gtgtgtgt 5 Adversarial Search and Games ...
gt gt gt gt gt gt
gtgtgtgt 6 Constraint Satisfaction Problems ...
gt gt gt gt gt
III Knowledge, reasoning, and planning
gt gt gt gt
gtgtgtgt 7 Logical Agents ...
gt gt gt gt
gtgtgtgt 8 First-Order Logic ...
gt gt gt gt
gtgtgtgt 9 Inference in First-Order Logic ...
gt gt gt gt gt
gtgtgtgt 10 Knowledge Representation ...
gt gt gt gt
gtgtgtgt 11 Automated Planning ...
gt gt gt gt
IV Uncertain knowledge and reasoning
gt gt gt gt
gtgtgtgt 12 Quantifying Uncertainty ...
gt gt gt gt
gtgtgtgt 13 Probabilistic Reasoning ...
gt gt gt gt
gtgtgtgt 14 Probabilistic Reasoning over Time ...
gt gt gt gt gt gt
gtgtgtgt 15 Probabilistic Programming ...
gt gt gt gt
gtgtgtgt 16 Making Simple Decisions ...
gt gt gt gt gt
gtgtgtgt 17 Making Complex Decisions ...
gt gt gt gt gt
gtgtgtgt 18 Multiagent Decision Making ...
gt gt gt gt gt
V Machine Learning
gt gt
, 19 Learning from Examples ...
gtgtgtgt gt gt gt gt gt
20 Learning Probabilistic Models ...
gtgtgtgt gt gt gt gt gt
21 Deep Learning ...
gtgtgtgt gt gt gt gt
22 Reinforcement Learning ...
gtgtgtgt gt gt gt gt
VI Communicating, perceiving, and acting
gt gt gt gt
23 Natural Language Processing ...
gtgtgtgt gt gt gt gt gt
24 Deep Learning for Natural Language Processing ...
gtgtgtgt gt gt gt gt gt gt gt gt
25 Computer Vision ...
gtgtgtgt gt gt gt gt
26 Robotics ...
gtgtgtgt gt gt gt
VII Conclusions
gt
27 Philosophy, Ethics, and Safety of AI ...
gtgtgtgt gt gt gt gt gt gt gt gt
28 The Future of AI
gtgtgtgt gt gt gt gt
, EXERCISES g t gt
1
INTRODUCTION
Notegtthatgtforgtmanygtofgtthegtquestionsgtingtthisgtchapter,gtwegtgivegtreferencesgtwheregtanswersgtc
angtbegtfoundgtrathergtthangtwritinggtthemgtout—thegtfullgtanswersgtwouldgtbegtfargttoogtlong.
1.1 What Is AI?
g t gt gt
Exercisegt1.1.#DEFA
Definegtingtyourgtowngtwords:g t (a)gtintelligence,gt(b)gtartificialgtintelligence,gt(c)gtagent,gt(d)gtr
a-gttionality,gt(e)gtlogicalgtreasoning.
a. Dictionarygtdefinitionsgtofgtintelligencegttalkgtaboutgt“thegtcapacitygttogtacquiregtandgtapp
lygtknowledge”gtorgt“thegtfacultygtofgtthoughtgtandgtreason”gtorgt“thegtabilitygttogtcompreh
endgtandgtprofitgtfromgtexperience.”g t Thesegtaregtallgtreasonablegtanswers,gtbutgtifgtwegtw
antgtsomethinggtquantifiablegtwegtwouldgtusegtsomethinggtlikegt“thegtabilitygttogtactgtsucces
sfullygtacrossgtagtwidegtrangegtofgtobjectivesgtingtcomplexgtenvironments.”
b. Wegtdefinegtartificialgtintelligencegtasgtthegtstudygtandgtconstructiongtofgtagentgtprogram
sgtthatgtperformgtwellgtingtagtgivengtclassgtofgtenvironments,gtforgtagtgivengtagentgtarchitect
ure;gttheygtdogtthegtrightgtthing.g t Angtimportantgtpartgtofgtthatgtisgtdealinggtwithgtthegtuncer
taintygtofgtwhatgtthegtcurrentgtstategtis,gtwhatgtthegtoutcomegtofgtpossiblegtactionsgtmightgtb
e,gtandgtwhatgtisgtitgtthatgtwegtreallygtdesire.
c. Wegtdefinegtangtagentgtasgtangtentitygtthatgttakesgtactiongtingtresponsegttogtperceptsgtfromgtan
gtenvi-gtronment.
d. Wegtdefinegtrationalitygtasgtthegtpropertygtofgtagtsystemgtwhichgtdoesgtthegt“rightgtthing”gt
givengtwhatgtitgtknows.g t SeegtSectiongt2.2gtforgtagtmoregtcompletegtdiscussion.g t Thegtbasi
cgtconceptgtisgtperfectgtrationality;gtSectiongt??gtdescribesgtthegtimpossibilitygtofgtachieving
gtperfectgtrational-gtitygtandgtproposesgtangtalternativegtdefinition.
e. Wegtdefinegtlogicalgtreasoninggtasgtthegtagtprocessgtofgtderivinggtnewgtsentencesgtfromgtold,gt
suchgtthatgtthegtnewgtsentencesgtaregtnecessarilygttruegtifgtthegtoldgtonesgtaregttrue.gt(Noticegtth
atgtdoesgtnotgtrefergttogtanygtspecificgtsyntaxgtorgtformalgtlanguage,gtbutgtitgtdoesgtrequiregtagtwel
l-definedgtnotiongtofgttruth.)
Exercisegt1.1.#TURI
ReadgtTuring’sgtoriginalgtpapergtongtAIgt(Turing,gt1950).g t Ingtthegtpaper,gthegtdiscussesgtsevera
lgtobjectionsgttogthisgtproposedgtenterprisegtandgthisgttestgtforgtintelligence.gtWhichgtobjectionsgtstillgtca
rry
©gt2023gtPearsongtEducation,gtHoboken,gtNJ.gtAllgtrightsgtres
erved.
