Upcoming SlideShare
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

# Saving this for later?

### Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime - even offline.

Standard text messaging rates apply

# AI

122
views

Published on

Published in: Education, Technology, Spiritual

0 Likes
Statistics
Notes
• Full Name
Comment goes here.

Are you sure you want to Yes No
• Be the first to comment

• Be the first to like this

Views
Total Views
122
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
3
0
Likes
0
Embeds 0
No embeds

No notes for slide
• But there’s still a fear: how would we eat?
• Picture from http://www.fourmilab.ch/babbage/hpb1910.html
• http://cogsci.ucsd.edu/~asaygin/tt/ttest.html http://www.loebner.net/Prizef/loebner-prize.html
• Herbert A. Simon and Allen Newell, &quot;Heuristic Problem Solving: The Next Advance in Operations Research,&quot; Operations Research, January-February 1958, 1-10.
• http://www.jabberwacky.com 2003 Loebner bot winner, beaten by two people
• http://www.alicebot.org/
• Note here that the language processing seems fine. It’s a lack of specific knowledge that is killing Alice.
• Is this AI? Is it just search? Is chess a representative “intelligent problem”? What about role playing games? Need a lot more knowledge.
• Knowledge – more later Perception – more later
• http://sern.ucalgary.ca/courses/CPSC/533/W99/presentations/L1_5A_Szuch_Boyd/game.html
• Parses: read bottom up. It gets worse at the end. 8064 total. With huge branches: tree, ground, dragging, wheel, wagon, girl, boy and girl, saw (8) under a tree can attach to ground, dragging, wheel, wagon, girl, boy and girl, saw. (7) On the ground (6) Dragging: wheel, wagon, girl, boy and girl (4) And: (one blue and one white) wheel, one blue and (one white wheel) (2) With a red wagon: girl, boy and girl, saw (3) How hard to check each? Current approaches: use statistics to guess right a lot of the time. But note that this one is ambiguous even for people.
• From Church and Patil, 1982, via http://www.cogs.susx.ac.uk/lab/nlp/gazdar/teach/nlp/nlpnode11.html Binomial coefficients: (a b) = a!/(b!(a-b)!) Cat(n) is the number of ways to parenthesize an expresssion of length n with two conditions: 1) Equal number of open and closes. That’s the first term. 2) They must be properly nested. That’s the second term, so it subtracts out the improper ones.
• Maybe the problem is representation. Can we punt search if we have trained a neural net to simply do the right thing? Maybe but now we have to search in a different space, the space of all neural nets.
• Neural net picture from http://hem.hj.se/~de96klda/NeuralNetworks.htm#2.1.2%20The%20Artificial%20Neuron
• http://diwww.epfl.ch/mantra/tutorial/english/mcpits/html/
• Note: They’re formal. Don’t need a lot of messy knowledge.
• Pictures from M. Muller, Computer Go, AIJ v.134, 1-2.
• Pictures from M. Muller, Computer Go, AIJ v.134, 1-2.
• Two beagles spot fire. Their barking alerts neighbors who call the police.
• http://www.jfsowa.com/pubs/semnet.htm
• http://www.rci.rutgers.edu/~cfs/472_html/Learn/Winston.html
• http://www.epub.org.br/cm/n09/historia/documentos_i.htm
• http://www.frc.ri.cmu.edu/~hpm/book97/index.html
• http://www.frc.ri.cmu.edu/~hpm/book98/fig.ch2/p027.html
• http://www.frc.ri.cmu.edu/~hpm/book97/ch2/index.html
• http://www.ai.mit.edu/people/brooks/papers/fast-cheap.pdf
• http://www.ri.cmu.edu/projects/project_163.html
• http://marsrovers.jpl.nasa.gov/gallery/artwork/rover2browse.html
• Sandstorm is a modified Humvee. It cost \$3M. The prize for the grand challenge is \$1M. Sandstorm made it 9 miles. 15 teams entered. Sandstorm got the farthest of all: At mile 7.4, on switchbacks in a mountainous section, vehicle went off course, got caught on a berm and rubber on the front wheels caught fire, which was quickly extinguished. Vehicle was command-disabled.
• Doctors will write the equivalent of 10 prescriptions for every person in the country, and as the population ages and drugs are more aggressively marketed, the numbers are growing. A shortage of pharmacists needed to keep up with the demand hit its peak two years ago, but it still exists in some areas. There are about 5,000 open pharmacy positions nationwide. And because of the paperwork and overwork associated with the enormous volume of prescriptions, mistakes still are being made. Most estimates put the number of mistakes at about 4 percent. Still, that represents more than 100-million mistakes a year - sometimes with tragic consequences.
• http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2001/artificial_intelligence/1531432.stm
• Is means equals. NPs are numbers.
• What do you want to do with your life?
• Why build machines to play games? That’s what we want to do. But what about knowledge-intensive yet boring tasks? BabelFish
• Using BabelFish babelfish.altavista.com Going both ways is particularly difficult since the first pass may create some nonstandard prose, which may make sense to people but be really hard to translate.
• What does bass mean? What role is the with phrase playing? To what does it attach? (see next slide)
• With is the instrument and attaches to caught
• With is an attribute and attaches see next slide to bass
• With is an attribute and attaches to bass
• Now the meaning of bass changes.
• But, if we’re speaking, how do we differentiate these two?
• Assuming we don’t mean Olive Oil
• Assuming we don’t mean Olive Oil
• Jacket used for riding
• What’s the weather like in Austin? Who invented the computer? Will computers ever be smarter than people?
• http://www.robotics.utexas.edu/rrg/learn_more/history/#firstrobot http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2001/artificial_intelligence/1531432.stm (UN estimate)
• A Commemorative Certificate for ascending the great wall in China. (input) I have ascended the great wall (output)
• ### Transcript

• 1. Artificial IntelligenceOur Attempt to Build Models of Ourselves
• 2. One Vision of an AI
• 3. A Calmer Vision
• 4. Could AI Stop This?
• 5. What is Artificial Intelligence?A.I. is the study of how to make computers do things atwhich, at the moment, people are better.
• 6. Or, Stepping Back Even Farther, Can WeBuild Artificial People?•Historical attempts•The modern quest for robots and intelligent agents•Us vs. Them
• 7. Historical Attempts - FrankensteinFrankenstein creates the fiend - illustration byBernie Wrightson (© 1977)The original story,published by MaryShelley, in 1818,describes the attemptof a true scientist,Victor Frankenstein,to create life.http://members.aon.at/frankenstein/frankenstein-novel.htm
• 8. Historical Attempts – The Turkhttp://www.theturkbook.com
• 9. Historical Attempts - EuphoniaJoseph Fabers Amazing Talking Machine (1830-40s). The Euphonia and other earlytalking devices are described in detail in a paper by David Lindsay called "Talking Head",Invention & Technology, Summer 1997, 57-63.Fromhttp://www.haskins.yale.edu/haskins/HEADS/SIMULACRA/euphonia.htmlAbout this device, Lindsay writes:It is "... a speech synthesizervariously known as the Euphonia andthe Amazing Talking Machine. Bypumping air with the bellows ... andmanipulating a series of plates,chambers, and other apparatus(including an artificial tongue ... ),the operator could make it speak anyEuropean language. A Germanimmigrant named Joseph Faber spentseventeen years perfecting theEuphonia, only to find when he wasfinished that few people cared."
• 10. Historical Attempts - RUR"CHEAP LABOR. ROSSUMS ROBOTS." "ROBOTS FOR THE TROPICS.  150 DOLLARS EACH.""EVERYONE SHOULD BUY HIS OWN ROBOT." "DO YOU WANT TO CHEAPEN YOUR OUTPUT?  ORDER ROSSUMS ROBOTS" In 1921, the Czech author Karel Capek produced the play R.U.R.(Rossums Universal Robots).http://www.maxmon.com/1921ad.htmSome references state that term "robot" was derived from the Czech wordrobota, meaning "work", while others propose that robota actually means "forcedworkers" or "slaves." This latter view would certainly fit the point that Capek wastrying to make, because his robots eventually rebelled against their creators, ranamok, and tried to wipe out the human race. However, as is usually the casewith words, the truth of the matter is a little more convoluted. In the days whenCzechoslovakia was a feudal society, "robota" referred to the two or three daysof the week that peasants were obliged to leave their own fields to work withoutremuneration on the lands of noblemen. For a long time after the feudal systemhad passed away, robota continued to be used to describe work that one wasntexactly doing voluntarily or for fun, while todays younger Czechs and Slovakstend to use robota to refer to work that’s boring or uninteresting.
• 11. The Roots of Modern Technology5thc B.C. Aristotelian logic invented1642 Pascal built an adding machine1694 Leibnitz reckoning machine
• 12. The Roots, continued1834 Charles Babbage’sAnalytical EngineAda writes of the engine, “TheAnalytical Engine has nopretensions whatever to originateanything. It can do whatever weknow how to order it to perform.”The picture is of a model built in the late 1800s by Babbage’s sonfrom Babbage’s drawings.
• 13. The Roots: Logic1848 George Boole The Calculus of Logicchocolatenutsmintchocolate and ¬ nuts and mint
• 14. Mathematics in the Early 20thCentury –(Looking Ahead: Will Logic be the Key toThinking?)1900 Hilbert’s program and the effort to formalizemathematics1931 Kurt Gödel’s paper, On Formally UndecidablePropositions1936 Alan Turing’s paper, On Computable Numbers with anapplication to the Entscheidungs problem
• 15. The Advent of the Computer1945 ENIAC The first electronic digital computer
• 16. 1949 EDVACThe first storedprogram computer
• 17. The Dartmouth Conference and the NameArtificial IntelligenceJ. McCarthy, M. L. Minsky, N. Rochester, and C.E.Shannon. August 31, 1955. "We propose that a 2month, 10 man study of artificial intelligence becarried out during the summer of 1956 atDartmouth College in Hanover, New Hampshire.The study is to proceed on the basis of theconjecture that every aspect of learning or anyother feature of intelligence can in principle beso precisely described that a machine can bemade to simulate it."
• 18. Time Line – The Big Picture50 60 70 80 90 00 101956 Dartmouth conference.1981 Japanese Fifth Generation project launched as theExpert Systems age blossoms in the US.1988 AI revenues peak at \$1 billion. AI Winter begins.academic \$ academic and routine
• 19. How Will We Recognize AI?1950 Alan Turing’s paper, Computing Machinery andIntelligence, described what is now called “The TuringTest”. http://www.abelard.org/turpap/turpap.htm1990 Loebner Prize established. Grand Prize of\$100,000 and a Gold Medal for the first computer whoseresponses are indistinguishable from a human.
• 20. The Origins of AI Hype1950 Turing predicted that in about fifty years "an averageinterrogator will not have more than a 70 percent chance ofmaking the right identification after five minutes ofquestioning".1957 Newell and Simon predicted that "Within ten years acomputer will be the worlds chess champion, unless the rulesbar it from competition."
• 21. Was Turing Right?Jabberwacky: Have you been to the Moon?User: No. Few have. Have you?Jabberwacky: Many a time my dear chap!User: What was the most interesting thing you saw?Jabberwacky: Fred.User: Why did you find him interesting?Jabberwacky: It just came into my head.User: What sorts of things do you generally find interesting?Jabberwacky: Nothing but smelly socks.A conversation with Jabberwacky, 2003
• 22. Another OneA Conversation with Alice, 2004>Whats the best place to eat Chinese food?ALICE: My favorite places are California and Maine. Hollandis cool too.>What do you like about Maine?ALICE: The special effects are the best.http://www.alicebot.org/
• 23. Alice, Again>Who invented paper towels?ALICE: The Chinese, long before Europeans.>Who invented pet rocks?ALICE: The Wright Brothers.>Who invented Barbie dolls?ALICE: Thomas Edison.
• 24. Were Newell and Simon Right?•1950 Claude Shannon published a paper describing howa computer could play chess.•1957 Newell and Simon predicted that a computer willbeat a human at chess within 10 years.•1967 MacHack was good enough to achieve a class-Crating in tournament chess.
• 25. Chess TodayIn 1997, Deep Blue beat GaryKasparov.
• 26. Why Did They Get it Wrong?They failed to understand at least three key things:•The need for knowledge (lots of it)•Scalability and the problem of complexity and exponentialgrowth•The need to perceive the world
• 27. ScalabilitySolving hard problemsrequires search in a largespace.To play master-levelchess requires searchingabout 8 ply deep. Soabout 358or 2⋅1012nodesmust be examined.
• 28. Exponential Growth
• 29. But Chess is Easy•The rules are simple enough to fit on one page•The branching factor is only 35.
• 30. A Harder OneJohn saw a boy and a girl with a red wagon with one blue andone white wheel dragging on the ground under a tree with hugebranches.
• 31. How Bad is the Ambiguity?•Kim (1)•Kim and Sue (1)•Kim and Sue or Lee (2)•Kim and Sue or Lee and Ann (5)•Kim and Sue or Lee and Ann or Jon (14)•Kim and Sue or Lee and Ann or Jon and Joe (42)•Kim and Sue or Lee and Ann or Jon and Joe or Zak (132)•Kim and Sue or Lee and Ann or Jon and Joe or Zak and Mel (469)•Kim and Sue or Lee and Ann or Jon and Joe or Zak and Mel or Guy (1430)•Kim and Sue or Lee and Ann or Jon and Joe or Zak and Mel or Guy and Jan(4862)The number of parses for an expression with n terms is the n’th Catalan number:−−=122)(nnnnnCat
• 32. Can We Get Around the Search Problem ?
• 33. How Much Compute Power Does it Take?From Hans Moravec, Robot Mere Machine to Transcendent Mind 1998.
• 34. How Much Compute Power is There?From Hans Moravec, Robot Mere Machine to Transcendent Mind 1998.
• 35. Evolution of the Main Ideas•Wings or not?•Games, mathematics, and other knowledge-poor tasks•The silver bullet?•Knowledge-based systems•Hand-coded knowledge vs. machine learning•Low-level (sensory and motor) processingand the resurgence of subsymbolic systems•Robotics•Natural language processing•Programming languages•Cognitive modeling
• 36. Symbolic vs. Subsymbolic AISubsymbolic AI: Modelintelligence at a level similar tothe neuron. Let such things asknowledge and planning emerge.Symbolic AI: Model suchthings as knowledge andplanning in data structures thatmake sense to theprogrammers that build them.(blueberry (isa fruit)(shape round)(color purple)(size .4 inch))
• 37. The Origins of Subsymbolic AI1943 McCulloch and Pitts A Logical Calculus of the IdeasImmanent in Nervous Activity“Because of the “all-or-none” character of nervousactivity, neural events and the relations among them canbe treated by means of propositional logic”
• 38. Interest in Subsymbolic AI40 50 60 70 80 90 00 10
• 39. Low-level (Sensory and Motor) Processingand the Resurgence of Subsymbolic Systems•Computer vision•Motor control•Subsymbolic systems perform cognitive tasks•Detect credit card fraud•The backpropagation algorithm eliminated a formalweakness of earlier systems•Neural networks learn.
• 40. The Origins of Symbolic AI•Games•Theorem proving
• 41. Games•Chess•Checkers:•1952-1962 Art Samuel built the first checkersprogram•Chinook became the world checkers champion in1994•Othello:•Logistello beat the world champion in 1997
• 42. Games•Chess•Checkers: Chinook became the world checkers champion in1994•Othello: Logistello beat the world champion in 1997•Role Playing Games: now we need knowledge•Go:
• 43. Mathematics1956 Logic Theorist (the first running AI program?)1961 SAINT solved calculus problems at the collegefreshman level1967 MacsymaGradually theorem proving has become well enoughunderstood that it is usually no longer considered AI1996 J Moore and others verified the correctness of theAMD5k86 Floating-Point Division algorithm
• 44. The Silver Bullet?Is there an “intelligence algorithm”?1957 GPS (General Problem Solver)Start Goal
• 45. But What About Knowledge?•Why do we need it?•How can we represent it and use it?•How can we acquire it?Find me stuff about dogs who save people’s lives.Around midnight, two beagles spotted afire in the house next door. Theirbarking alerted their owners, who called911.
• 46. Representing Knowledge - Logic1958 McCarthy’s paper, “Programs with Common Sense”at(I, car) ⇒ can (go(home, airport, driving))at(I, desk) ⇒ can(go(desk, car, walking))1965 Resolution theorem proving invented
• 47. Representing Knowledge- Semantic Nets1961
• 48. Representing Knowledge – CapturingExperienceRepresenting Experience with Scripts, Frames, and Cases1977 ScriptsJoe went to a restaurant. Joe ordered a hamburger. When thehamburger came, it was burnt to a crisp. Joe stormed outwithout paying.The restaurant script:Did Joe eat anything?
• 49. Representing Knowledge - RulesExpert knowledge in many domains can be captured inrules.From XCON (1982):If: the most current active context is distributingmassbus devices, andthere is a single-port disk drive that has not beenassigned to a massbus, andthere are no unassigned dual-port disk drives, andthe number of devices that each massbus should support is known,andthere is a massbus that has been assigned at least one disk drive thatshould support additional disk drives, andthe type of cable needed to connect the disk drive to the previousdevice on the massbus is knownThen: assign the disk drive to the massbus.
• 50. Representing Knowledge – Probabilistically1975 Mycin attaches probability-like numbers to rules1970s Probabilistic models of speech recognition1980s Statistical Machine Translation systems1990s large scale neural netsIf: (1) the stain of the ogranism is gram-positive, and(2) the morphology of the organism is coccus, and(3) the growth conformation of the organism is clumpsThen: there is suggestive evidence (0.7) that the identity ofthe organism is stphylococcus.
• 51. The Rise of Expert Systems1967 Dendral – a rule-based system that inferedmolecular structure from mass spectral and NMR data1975 Mycin – a rule-based system to recommendantibiotic therapy1975 Meta-Dendral learned new rules of massspectrometry, the first discoveries by a computer to appear ina refereed scientific journal1979 EMycin – the first expert system shell1980’s The Age of Expert Systems
• 52. Expert Systems – The Heyday1979 Inference1980 IntelliCorp1981 Teknowledge1983 Carnegie Group1980 XCON (R1) – first real commercial expert system atDEC, configures VAX systems1981 Japanese Fifth Generation project launched as theExpert Systems age blossoms in the US.1984 Gold Hill Common Lisp1986 neural net start up companies appear1988 AI revenues peak at \$1 billion. AI Winter begins.Selling expert system shells
• 53. Expert Systems – Today: MedicineOne example domain, medicine, has expert systems whosetasks include:•arrhythmia recognition from electrocardiograms•coronary heart disease risk group detection•monitoring the prescription of restricted use antibiotics•early melanoma diagnosis•gene expression data analysis of human lymphoma•breast cancer diagnosis
• 54. Expert Systems – Today: Build Your OwneXpertise2Go: http://www.expertise2go.com/AcquiredIntelligence: http://www.aiinc.ca/demos/(whales, graduate school)
• 55. Taking the SAT ExamQUESTION: Sodium azide is used in air bags to rapidly producegas to inflate the bag. The products of the decomposition reactionare:(a) Na and water.(b) Ammonia and sodium metal.(c) N2 andO2(d) Sodium and nitrogen gas.(e) Sodium oxide and nitrogengas.(d) Sodium and nitrogen gas.
• 56. SYSTEM-GENERATED ANSWER:# What are the products of the given decomposition reaction?•By definition, oxidation-reduction reactions occur when electrons aretransferred from the atom that is oxidized to the atom that is reduced. We need tolook for changes in the oxidation states of the elements in the reaction.•In the reactants, the oxidation state(s) of the element Na is/are (1). In theproduct, the oxidation state(s) is/are (0).Therefore, the reaction causes a change in oxidation state.•Therefore, this is an oxidation reduction reaction.•By definition, a Binary Ionic-Compound Decomposition Reaction occurs whena binary ionic compound is heated.•Therefore, this reaction is a Binary-Ionic Compound Decomposition reaction.•In general, a Binary Ionic-Compound Decomposition Reaction converts abinary ionic-compound into basic elements.•In this reaction, NaN3 reacts to produce Na and N2.# The products of the decomposition reaction are:(d) Sodium and nitrogen gas.The work of Bruce Porter et al here at UT
• 57. What About Things that People Do Easily?•Common sense•Moving Around•Language
• 58. What About Things that People Do Easily?•Common sense•CYC•UT (http://www.cs.utexas.edu/users/mfkb/RKF/tree/ )•WordNet (http://www.cogsci.princeton.edu/~wn/)•Moving around•Language
• 59. Hand-Coded Knowledge vs. Machine Learning•How much work would it be to enter knowledge by hand?•Do we even know what to enter?1952-62 Samuel’s checkers player learned its evaluationfunction1975 Winston’s system learned structural descriptionsfrom examples and near misses1984 Probably Approximately Correct learning offers atheoretical foundationmid 80’s The rise of neural networks
• 60. Robotics - Tortoise1950 W. Grey Walter’s light seeking tortoises. In thispicture, there are two, each with a light source and a lightsensor. Thus they appear to “dance” around each other.
• 61. Robotics – Hopkins Beast1964 Two versions of the Hopkins beast, which used sonar toguide it in the halls. Its goal was to find power outlets.
• 62. Robotics - Shakey1970 Shakey (SRI)was driven by a remote-controlled computer,which formulated plansfor moving and acting.It took about half anhour to move Shakeyone meter.
• 63. Robotics – Stanford Cart1971-9 Stanford cart.Remote controlled byperson or computer.1971 follow the whiteline1975 drive in a straightline by tracking skyline1979 get throughobstacle courses. Cross30 meters in five hours,getting lost one timeout of four
• 64. Planning vs. ReactingIn the early days: substantial focus on planning (e.g., GPS)1979 – in “Fast, Cheap and Out of Control”, RodneyBrooks argued for a very different approach. (No, I’m nottalking about the 1997 movie.)http://www.ai.mit.edu/people/brooks/papers/fast-cheap.pdfhttp://www.ai.mit.edu/projects/ants/The Ant, has 17 sensors.They are designed to workin colonies.
• 65. Robotics - Dante1994 Dante II (CMU) exploredthe Mt. Spurr (Aleutian Range,Alaska) volcano. High-temperature, fumarole gassamples are prized by volcanicscience, yet their sampling posessignificant challenge. In 1993,eight volcanologists were killedin two separate events whilesampling and monitoringvolcanoes.Using its tether cable anchored at the crater rim, Dante II is ableto descend down sheer crater walls in a rappelling-like manner togather and analyze high temperature gasses from the crater floor.
• 66. Robotics - Sojournerhttp://antwrp.gsfc.nasa.gov/apod/ap991030.htmlOct. 30, 1999 Sojourner on Mars. Powered by a 1.9 square footsolar array, Sojourner can negotiate obstacles tilted at a 45degree angle. It travels at less than half an inch per second.
• 67. Robotics – Mars RoverTutorial on Rover:http://marsrovers.jpl.nasa.gov/gallery/video/animation.html
• 68. SandstormMarch 13, 2004 - A DARPA Grand Challenge: an unmannedoffroad race, 142 miles from Barstow to Las Vegas.
• 69. Moving Around and Picking Things UpPhil, the drug robot, introduced in 2003
• 70. Robotics - Aibo1999 Sony’s Aibo petdog
• 71. What Can You Do with an Aibo?1997 – First official Rob-Cup soccer matchPicture from 2003competition
• 72. Robotics - Coghttp://www.eecs.mit.edu/100th/images/Brooks-Cog-Kismet.html1998 – now CogHumanoidintelligence requireshumanoidinteractions with theworld.
• 73. At the Other End of the Spectrum - Roomba2001 A robotvacuum cleaner
• 74. Natural Language Processing1964 STUDENT solves algebra word problemsThe distance from New York to Los Angeles is 3000 miles. If theaverage speed of a jet plane is 600 miles per hour, find the time ittakes to travel from New York to Los Angeles by jet.1965 ELIZA models a Rogerian therapistyoung woman: Men are all alike.eliza: In what way?young woman: Theyre always bugging us about somethingspecific or other.eliza: Can you think of a specific example?young woman: Well, my boyfriend made me come here.eliza: Your boyfriend made you come here?
• 75. NLP, continued1966 Alpac report kills work on MT1971 SHRDLU
• 76. NLP, continued1973 Schank – a richer limited domain: children’s storiesSuzie was invited to Mary’s birthday party. She knewshe wanted a new doll so she got it for her.1977 Schank – scripts add a knowledge layer – restaurantstories1970’s and 80’s sophisticated grammars and parsersBut suppose we want generality? One approach is “shallow”systems that punt the complexities of meaning.
• 77. NLP Today•Grammar and spelling checkers•Spelling: http://www.spellcheck.net/•Chatbots•See the list at:http://www.aaai.org/AITopics/html/natlang.html#chat/•Speech systems•Synthesis: The IBM system:•http://www.research.ibm.com/tts/coredemo.html
• 78. Machine Translation: An Early NLApplication1949 Warren Weaver’s memo suggesting MT1966 Alpac report kills government fundingEarly 70s SYSTRAN develops direct Russian/English systemEarly 80s knowledge based MT systemsLate 80s statistical MT systems
• 79. MT TodayAustin Police are trying to find the person responsible for robbing abank in Downtown Austin.El policía de Austin está intentando encontrar a la personaresponsable de robar un banco en Austin céntrica.The police of Austin is trying to find the responsible person to rob abank in centric Austin.
• 80. MT TodayA Florida teen charged with hiring an undercover policeman toshoot and kill his mother instructed the purported hitman not todamage the family television during the attack, police said onThursday.Un adolescente de la Florida cargado con emplear a un policía dela cubierta interior para tirar y para matar a su madre mandó ahitman pretendida para no dañar la televisión de la familia duranteel ataque, limpia dicho el jueves.An adolescent of Florida loaded with using a police of the innercover to throw and to kill his mother commanded to hitman tried notto damage the television of the family during the attack, clean saidThursday.
• 81. MT TodayI have a dream, that my four little children will one day live in anation where they will not be judged by the color of their skin butby the content of their character. I have a dream today – MartinLuther KingI am a sleepy, that my four small children a day of alive in anation in where they will not be judged by the color of its skin butby the content of its character. I am a sleepy today. (Spanish)http://www.shtick.org/Translation/translation47.htm
• 82. Why Is It So Hard?Sue caught the bass with her new rod.
• 83. Why Is It So Hard?Sue caught (the bass) (with her new rod).
• 84. Why Is It So Hard?Sue caught the bass with the dark stripes.
• 85. Why Is It So Hard?Sue caught (the bass with the dark stripes).
• 86. Why Is It So Hard?Sue played the bass with her new bow.
• 87. Why Is It So Hard?Sue played the bass with her new bow.Sue played the bass with her new beau.
• 88. Why Is It So Hard?Olive oil
• 89. Why Is It So Hard?Olive oil
• 90. Why Is It So Hard?Peanut oil
• 91. Why Is It So Hard?Coconutoil
• 92. Why Is It So Hard?Babyoil
• 93. Why Is It So Hard?Cooking oil
• 94. MT TodayIs MT an “AI complete” problem?•John saw a bicycle in the store window. He wanted it.•John saw a bicycle in the store window. He pressed hisnose up against it.•John saw the Statue of Liberty flying over New York.•John saw a plane flying over New York.
• 95. Text Retrieval and Extraction•Try Ask Jeeves: http://www.askjeeves.com•To do better requires:•Linguistic knowledge•World knowledge•Newsblaster: http://www1.cs.columbia.edu/nlp/newsblaster/
• 96. Programming Languages1958 Lisp – a functional programming language with asimple syntax.1972 PROLOG - a logic programming language whoseprimary control structure is depth-first searchancestor(A,B) :- parent(A,B)ancestor(A,B) :- parent(A,P), ancestor(P,B)1988 CLOS (Common Lisp Object Standard) published.Draws on ideas from Smalltalk and semantic nets(successor SitA ActionP)
• 97. Cognitive ModelingSymbolic Modeling1957 GPS1983 SOARNeuron-Level ModelingMcCulloch Pitts neurons: all or none responseMore sophisticated neurons and connectionsMore powerful learning algorithm
• 98. Making Money – Software•Expert systems to solve problems in particular domains•Expert system shells to make it cheaper to build new systemsin new domains•Language applications•Text retrieval•Machine Translation•Text to speech and speech recognition•Data mining
• 99. Making Money - Hardware1980 Symbolics founded1986 Thinking Machines introduces the Connection Machine1993 Symbolics declared bankruptcySymbolics 3620 System c 1986:Up to 4 Mwords (16 Mbytes)optional physical memory, one190 Mbyte fixed disk, integralEthernet interface, five backplaneexpansion slots, options includean additional 190 Mbyte disk or1/4 tape drive, floating pointaccelerator, memory, RS232Cports and printers.
• 100. Making Money - Robots1962 Unimation, first industrialrobot company, founded. Sold adie casting robot to GM.1990 iRobot founded, a spinoffof MIT2000 The UN estimated thatthere are 742,500 industrial robotsin use worldwide. More than halfof these were being used in Japan.2001 iRobot markets Roombafor \$200.
• 101. The Differences Between Us and ThemEmotionsUnderstandingConsciousness
• 102. EmotionsThe robot Kismet shows emotionssad surprisehttp://www.ai.mit.edu/projects/humanoid-robotics-group/kismet/
• 103. UnderstandingSearle’s Chinese Room
• 104. ConsciousnessMe You
• 105. Today: The Difference Between Us and Them
• 106. Today: Computer as ArtistTwo paintings done by Harold Cohen’s Aaron program:
• 107. Why AI?"AI can have two purposes. One is to use the power ofcomputers to augment human thinking, just as we usemotors to augment human or horse power. Roboticsand expert systems are major branches of that. Theother is to use a computers artificial intelligence tounderstand how humans think. In a humanoid way. Ifyou test your programs not merely by what they canaccomplish, but how they accomplish it, they yourereally doing cognitive science; youre using AI tounderstand the human mind."- Herb Simon