CS440ECE448 Artificial Intelligence Section Q course website httpslazebnicsillinoisedufall17 What is AI What is AI Four possible definitions textbook ch 1 Thinking humanly ID: 624072
Download Presentation The PPT/PDF document "CS440/ECE448: Artificial Intelligence" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
CS440/ECE448: Artificial IntelligenceSlide2
CS440/ECE448: Artificial IntelligenceSection Q course website:
http://slazebni.cs.illinois.edu/fall17/Slide3
What is AI?Slide4
What is AI?Four possible definitions (textbook ch. 1):
Thinking
humanly
Thinking rationally
Acting humanlyActing
rationallySource: Berkeley CS188 materialsSlide5
AI definition 1: Thinking humanlyNeed to study the brain as an information processing machine: cognitive science and neuroscienceSlide6
AI definition 1: Thinking humanlyCan we build a brain?
Computers
BrainsDigital
AnalogFixed architectureEvolving architecture
Fixed processing speedNo system clockModular, (primarily) serialMassively parallel
Separate hardware, softwareNo distinction between hardware and softwareSeparate computation, memory
No distinction between computation and memory
DisembodiedEmbodied
http://
scienceblogs.com
/
developingintelligence
/2007/03/27/why-the-brain-is-not-like-a-co/Slide7
The Turing Test
What capabilities would a computer need to have to pass the Turing Test?
Natural language processingKnowledge representationAutomated reasoningMachine learning
Turing predicted that by the year 2000, machines would be able to fool 30% of human judges for five minutes
AI definition 2: Acting humanlyA. Turing, Computing machinery and intelligence, Mind 59, pp. 433-460, 1950Slide8
AI is solved?
http://
www.bbc.com
/news/technology-27762088
http://en.wikipedia.org
/wiki/Eugene_Goostmanhttp://
www.scottaaronson.com/blog/?p=1858Slide9
Variability in protocols, judgesSuccess depends on deception!
Chatbots can do well using “cheap tricks”
First example: ELIZA (1966)What’s wrong with the Turing test?Slide10
A better Turing test?Winograd schema: Multiple choice questions that can be easily answered by people but cannot be answered by computers using “cheap tricks”
The
trophy would not fit in the brown suitcase because it was so small. What was so small?The trophy
The brown suitcase
H. Levesque, On our best behaviour, IJCAI 2013
http://www.newyorker.com/online/blogs/elements/2013/08/why-cant-my-computer-understand-me.htmlSlide11
A better Turing test?Winograd schema: Multiple choice questions that can be easily answered by people but cannot be answered by computers using “cheap tricks”
The
trophy would not fit in the brown suitcase because it was so large. What was so large
?The trophyT
he brown suitcaseH. Levesque, On our best
behaviour, IJCAI 2013http://www.newyorker.com/online/blogs/elements/2013/08/why-cant-my-computer-understand-me.htmlSlide12
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
steel. What was made of steel?The large ballThe tableSlide13
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
styrofoam. What was made of styrofoam?The large ballThe tableSlide14
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
styrofoam. What was made of styrofoam?The large ballThe table
The sack of potatoes had been placed below the bag of flour, so it had to be moved first. What had to be moved first?
The sack of potatoesThe bag of flourSlide15
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
styrofoam. What was made of styrofoam?The large ballThe table
The sack of potatoes had been placed above
the bag of flour, so it had to be moved first. What had to be moved first?The sack of potatoesThe bag of flourSlide16
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
styrofoam. What was made of styrofoam?The large ballThe table
The sack of potatoes had been placed above
the bag of flour, so it had to be moved first. What had to be moved first?The sack of potatoesThe bag of flourSam tried to paint a picture of shepherds with sheep, but they ended up looking like
rabbits. What looked like rabbits?The shepherdsThe sheepSlide17
Winograd schema: More examplesThe large ball crashed right through the
table because it was made of
styrofoam. What was made of styrofoam?The large ballThe table
The sack of potatoes had been placed above
the bag of flour, so it had to be moved first. What had to be moved first?The sack of potatoesThe bag of flourSam tried to paint a picture of shepherds with sheep, but they ended up looking like
golfers. What looked like golfers?The shepherds
The sheepSlide18
Winograd schemaAdvantages over standard Turing test
Test can be administered and graded by machineDoes not depend on human subjectivity
Does not require ability to generate English sentencesQuestions cannot be evaded using verbal dodgesQuestions can be made “Google-proof” (at least for now…)
Winograd schema challengeHeld at IJCAI conference in July 2016
Six entries, best system got 58% of 60 questions correct (humans get 90% correct)Slide19
AI definition 3: Thinking rationallyIdealized or “right” way of thinkingLogic:
patterns of argument that always yield correct conclusions when supplied with correct premises
“Socrates is a man; all men are mortal; therefore Socrates is mortal.”Logicist approach to AI: describe problem in formal logical notation and apply general deduction procedures to solve it
Problems with the logicist approachComputational complexity of finding the solution
Describing real-world problems and knowledge in logical notationDealing with uncertaintyA lot of “rational” behavior has nothing to do with logicSlide20
AI definition 4: Acting rationallyA rational agent
acts to optimally achieve its goalsGoals are application-dependent and are expressed in terms of the
utility of outcomesBeing rational means maximizing your utility or expected utility in the presence of uncertainty
In practice, utility optimization is subject to the agent’s computational constraints (bounded rationality or bounded optimality)Slide21
Utility maximization formulationAdvantagesDefinition is about the agent’s decisions/actions, not the cognitive process behind them
Generality: goes beyond explicit reasoning, and even human cognition altogetherPracticality: can be adapted to many real-world problems
Naturally accommodates uncertaintyAmenable to good scientific and engineering methodologyAvoids philosophy and psychologyDisadvantages?It may be hard to formulate utility functions, especially for complex open-ended tasksThe AI may end up “gaming” the utility function, or its operation may have unintended consequences
Has limited applicability to humansSlide22
Humans vs. rationalitySlide23
Philosophy of this classLearn to program computers to solve hard problems traditionally thought to require human intelligence
Adopt the “rational agent” definition of AI Follow a sound scientific/engineering methodology
Consider limited application domainsUse well-defined input/output specifications Define operational criteria amenable to objective validationZero in on essential problem features
Focus on principles and basic building blocks