week 4

Slide1: 

COM1070: Introduction to Artificial Intelligence week 4 Prof. M. Walker Dr. G. Gorrell Computer Science Department University of Sheffield

AI In the News: 

AI In the News http://www.aaai.org/aitopics/html/current.html

Presentations—Grading Policy: 

Presentations—Grading Policy Content (50%) Clearly defined topic Well supported arguments Clear analysis Organization (25%) Introduction/Body/Conclusion Logical flow Presentation (25%) Clear language Timed well

What to think about when designing presentation: 

What to think about when designing presentation What have I learned about this topic? What is the current state of the art? Where is the interest in this topic (e.g., business, research, medicine) and why? What are some concrete examples of the technology? Where is the technology heading? What do I think about this technology? Does it work now? Why or why not? If yes, what was the key breakthrough? If not, what is needed? Can I imagine myself using this technology? Why or why not?

Review: 

Review AI Applications Knowledge Representation Inference NOW: Evaluation

Which evaluation measures?: 

Which evaluation measures? Intelligence Can Machines Think? (philosophical) What is intelligence? Believability When I am interacting with this agent do I ‘suspend disbelief’ and interact naturally? Usability Does this agent help with some real world task How does its human-like qualities help or hinder task accomplishment? Commercial Viability

Overview: 

Overview The Turing Test The Loebner Prize Current Examples Turing’s Analysis of Potential Criticisms Searle’s Chinese Room Historical Candidates for Turing Test ELIZA PARRY

The Turing Test: 

The Turing Test Turing in 1950 published a philosophical paper designed to stop people arguing about whether or not machines could think. Compare: Can machines think? Can machines fly? Can machines swim? He proposed that the question be replaced with a test, which is what is now called the Turing Test.

Turing’s proposed test was a variation on a common parlour game called the imitation game: 

Turing’s proposed test was a variation on a common parlour game called the imitation game An interrogator in another room asks questions of a subject by teletype(!), to guess their gender. The subject is sometimes a man and sometimes a woman. Turing suggested a computer could replace a subject and one could ask whether the interrogator would notice. If, after some agreed time, the interrogator cannot distinguish situations where a machine has been substituted for the man/woman, we should just agree to say the machine can think (says Turing).

Modern Day Turing Test: 

Modern Day Turing Test Your only interaction with a subject in a different physical location is by texting The subject is sometimes a computer and sometimes a person. Can you tell whether you are texting with a person or a computer?. If, after some agreed length of time, you cannot distinguish situations where a machine has been substituted for a person, we should just agree to say the machine can think (says Turing).

Slide11: 

  ‘I believe that in about fifty years’ time it will be possible to program computers….to make them play the imitation game so well that an average interrogator will not have more than 70 percent chance of making the right identification after five minutes of questioning…… I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.’ Turing 1950 .

Loebner Prize: 

Loebner Prize Competition started in 1991 by Hugh Loebner. http://www.loebner.net/Prizef/loebner-prize.html Judges type questions in at terminal: replied to either by human or a computer contestant. Rate responses according to how human-like they appear: Prize of 2000 dollars to most human-like. Grand prize of 100,000 dollars if a program does better than one of the humans.

1997 Winner: 

1997 Winner Catherine: Intelligent Research Ltd, Designed and largely programmed by members of Sheffield’s Natural Language Processing Group. First year of general conversation track

Catherine: 1997 Loebner Winner: 

Catherine: 1997 Loebner Winner P: Hi, my name is Catherine, what's yours? . J: I'm Jurgen. P: Well, it's nice to meet you, Jurgen. Did you see that story on CNN last night about the lesbian couple who came out at a White House party on Sunday? J: NO. I just came in yesterday. I'm still kind of jet-lagged. P: Ellen Degeneres was one of them - she was kissing her lover in front of Bill Clinton. Do you think he was very embarrassed? J: I don't know? Don't you think it's a little childish if the leader of the most powerful nation on the planet is embarrassed by seeing two lesbians kiss? P: Probably! No. – Are you still a Friend of Bill Clinton?

Slide15: 

J: Never met him. P: You really don't like the guy--well, at least that's definite--I thought most people were just indifferent or numb by now. J: Have you seen the movie Mars attacs? P: Give me a break -- I'm just here to answer dumb questions - not that! What were we talking about... Ok.. I guess that means you really think there's serious behind Whitewater, the Special prosecutor and all that stuff? J: ]I'm cold. P: But how do you actually feel about him, now in the number two term? J: Hey, don't you get it? I don't want to talk about Clinton.

Slide16: 

P: Sorry--my attention slipped for a moment--can we pushon to something more gripping. where was I.. oh yes OK, but what happens now--does it all drift on, with everyone waiting to see how much his friends have said about him under threat of lots and lots of jail-or will something dramatic actually happen? J: Couldn't care less. P: Well, you’re more concerned about him that I am, I suppose--I think we've had worse, and who knows what I might have agreed to in that locked hotel room. J: You mean you might have been thrilled to have sex with the President? P: sex, I knew it would come up sooner or later. Time to get off this one I think and onto something more down to earth! J: What kind of earth? P: Somebody once told me it's true. Like I was saying I suppose it depends on where you're coming from, but as the song goes in My Fair Lady, "Why can't a woman be more like a man? " J: @@04

Catherine (1) Knew Current Topics(2) Tried to Maintain Initiative: 

Catherine (1) Knew Current Topics (2) Tried to Maintain Initiative

2006 Loebner Winner: 

2006 Loebner Winner Most recent winner (2006) is Joan http://www.channel4.com/news/articles/world/joan%20programme%20wins%20ai%20prize/169180 2007 contest held yesterday http://loebner.net/Prizef/2007_Contest/Rules.html

Alice and her descendants: 

Alice and her descendants http://ccdevnet.org/www.knytetrypper.com/ain.html

Weak vs. Strong AI: 

Weak AI: computer is valuable tool for study of mind, ie. can formulate and test hypotheses rigorously. Strong AI: appropriately programmed computer really is a mind, can be said to understand, and has other cognitive states. Engineering: Who cares? Does it work? Weak vs. Strong AI

What measures matter?: 

What measures matter? Intelligence Can Machines Think? (philosophical) What is intelligence? Believability When I am interacting with this agent do I ‘suspend disbelief’ and interact naturally? Usability Does this agent help with some real world task How does its human-like qualities help or hinder task accomplishment? Commercial Viability

Usability and Commercial Viability: 

Usability and Commercial Viability Expressive interfaces the ‘appearance’ of an interface can elicit positive responses Negative aspects computers frustrate users Anthropomorphism and interface agents The pros and cons Designing synthetic characters

Expressive interfaces: 

Expressive interfaces Colour, icons, sounds, graphical elements and animations are used to make the ‘look and feel’ of an interface appealing Conveys an emotional state In turn this can affect the usability of an interface People are prepared to put up with certain aspects of an interface (e.g. slow download rate) if the end result is very appealing and aesthetic

Friendly interfaces: 

Friendly interfaces Microsoft pioneered friendly interfaces for technophobes - ‘At home with Bob’ software 3D metaphors based on familiar places (e.g. living rooms) Agents in the guise of pets (e.g. bunny, dog) were included to talk to the user Make users feel more at ease and comfortable

User frustration: 

User frustration Many causes: When an application doesn’t work properly or crashes When a system doesn’t do what the user wants it to do When a user’s expectations are not met When a system does not provide sufficient information to enable the user to know what to do When error messages pop up that are vague, obtuse or condemning When the appearance of an interface is garish, noisy, gimmicky or patronizing

Error messages: 

Error messages “The application Word Wonder has unexpectedly quit due to a type 2 error.” Why not instead: “the application has expectedly quit due to poor coding in the operating system”

Website error message…: 

Website error message…

More helpful error message: 

More helpful error message “The requested page /helpme is not available on the web server. If you followed a link or bookmark to get to this page, please let us know, so that we can fix the problem. Please include the URL of the referring page as well as the URL of the missing page. Otherwise check that you have typed the address of the web page correctly. The Web site you seek Cannot be located, but Countless more exist.”

Should computers say they’re sorry?: 

Should computers say they’re sorry? Reeves and Nass (1996) argue that computers should be made to apologize Should emulate human etiquette Would users be as forgiving of computers saying sorry as people are of each other when saying sorry? How sincere would they think the computer was being? For example, after a system crash: “I’m really sorry I crashed. I’ll try not to do it again” How else should computers communicate with users?

Anthropomorphism: 

Anthropomorphism Attributing human-like qualities to inanimate objects (e.g. cars, computers) Well known phenomenon in advertising Dancing butter, drinks, breakfast cereals Much exploited in human-computer interaction Make user experience more enjoyable, more motivating, make people feel at ease, reduce anxiety

Which do you prefer?: 

Which do you prefer? 1. As a welcome message “Hello Chris! Nice to see you again. Welcome back. Now what were we doing last time? Oh yes, exercise 5. Let’s start again.” “User 24, commence exercise 5.”

Which do you prefer? : 

Which do you prefer? 2. Feedback when get something wrong “Now Chris, that’s not right. You can do better than that.Try again.” “Incorrect. Try again.” Is there a difference as to what you prefer depending on type of message? Why?

Evidence to support anthropomorphism: 

Evidence to support anthropomorphism Reeves and Nass (1996) found that computers that flatter and praise users in education software programs -> positive impact on them “Your question makes an important and useful distinction. Great job!” Students were more willing to continue with exercises with this kind of feedback

Criticism of anthropomorphism: 

Criticism of anthropomorphism Deceptive, make people feel anxious, inferior or stupid People tend not to like screen characters that wave their fingers at the user & say: Now Chris, that’s not right. You can do better than that.Try again.” Many prefer the more impersonal: “Incorrect. Try again.” Studies have shown that personalized feedback is considered to be less honest and makes users feel less responsible for their actions (e.g. Quintanar, 1982)

Virtual characters: 

Virtual characters Increasingly appearing on our screens Web, characters in videogames, learning companions, wizards, newsreaders, popstars Provides a persona that is welcoming, has personality and makes user feel involved with them

Disadvantages: 

Disadvantages Lead people into false sense of belief, enticing them to confide personal secrets with chatterbots (e.g. Alice) Annoying and frustrating E.g. Clippy Not trustworthy virtual e-commerce assistants?

Miss boo.com: 

Miss boo.com What do you think of Miss boo?

Persuasive advice?: 

Persuasive advice?

Miss Boo tanked: 

Miss Boo tanked http://news.bbc.co.uk/1/hi/business/997446.stm

Virtual characters: agents: 

Virtual characters: agents Can be classified in terms of the degree of anthropomorphism they exhibit: Synthetic characters animated agents emotional agents embodied conversational agents

(i)Synthetic characters -Silas the dog: 

(i)Synthetic characters -Silas the dog (Blumberg, 1996 - MIT) • autonomous, with internal states and able to respond to external events

(ii) Animated agents: 

(ii) Animated agents Play a collaborative role at the interface Often cartoon-like e.g. Herman the bug (Lester et al, 1997 Intellimedia) flies into plants & explains things on-the-fly & gives advice to students

(iii) Emotional agents: 

(iii) Emotional agents Pre-defined personality and set of emotions that user can change The Woggles, Bates, 1994

(iv) Embodied conversational agents: 

(iv) Embodied conversational agents Rea, real-estate agent, showing user an apartment Human-like body Uses gesture, non-verbal communication (facial expressions, winks) while talking Sophisticated AI techniques used to enable this form of interaction Cassell, 2000, MIT

Conversation with Rea: 

Conversation with Rea Mike approaches screen and Rea turns to face him and says: Hello. How can I help you? Mike: I’m looking to buy a place near MIT. Rea nods, indicating she is following. Rea: I have a house to show you. (picture of a house appears on the screen) Rea: it is in Somerville. Mike: Tell me about it. Rea looks up and away while she plans what to say. Rea: It’s big. Rea makes an expansive gesture with her hands. Mike brings his hands up as if to speak, so Rea does not continue, waiting for him to speak. Mike: Tell me more about it. Rea: Sure thing. It has a nice garden...

Which is the most believable agent?: 

Which is the most believable agent? Believability refers to the extent to which users come to believe an agent’s intentions and personality Appearance is very important Are simple cartoon-like characters or more realistic characters, resembling the human form more believable? Behaviour is very important How an agent moves, gestures and refers to objects on the screen Exaggeration of facial expressions and gestures to show underlying emotions (cf animation industry)

Tactical Military Agentshttp://www.isi.edu/soar/soar-ifor-project.html: 

Tactical Military Agents http://www.isi.edu/soar/soar-ifor-project.html Goal-driven behavior Knowledge-intensive behavior Reactivity Real-time performance Conforming to human reaction times and limits Overlapping of performance of multiple high-level tasks Multi-agent coordination Communication Agent Modeling Temporal Reasoning Planning Maintaining Episodic Memory

Apache Helicopter Pilot: 

Apache Helicopter Pilot

Steve: An Embodied Intelligent Agent for Virtual Environments: 

Steve: An Embodied Intelligent Agent for Virtual Environments 3D agent that interacts with students in virtual environments Can take different roles: teammate, teacher, guide, demonstrator Multiple trainees and agents work together in virtual teams Intelligent tutoring in the context of a shared team environment

General Capabilities: 

General Capabilities Planning, replanning, and plan execution Path planning Collaborative, mixed initiative dialogue Student monitoring Question answering Control of a human figure

Steve trains student in virtual world: 

Steve trains student in virtual world Student is the Blue Avatar

Today’s thought question: 

Today’s thought question Does the Turing Test make sense as a way to evaluate AI technology in today’s world? Why?

The Turing Test?: 

The Turing Test?

Criticism of the Turing Test : 

Criticism of the Turing Test  There are at least two alternative positions which criticise AI with respect to the Turing Test: i. ‘Too hard’ definition of Artificial Intelligence. Computers not likely to be able to pass the test. ii. Hollow shell criticism. Computer may pass test, but computers still won’t be able to think. As we shall see (on i) computers aren’t doing badly and are getting better. On (ii) the answer just begs the question as to what thinking is--which was Turing’s point in the first place!

The Chinese Room: 

The Chinese Room Hollow Shell

Searle’s Chinese Room: 

Searle’s Chinese Room An operator O. sits in a room; Chinese symbols come in which O. does not understand. He has explicit instructions (a program!) in English in how to get an output stream of Chinese characters from all this, so as to generate “answers” from “questions”. But of course he understands nothing even though Chinese speakers who see the output find it correct and indistinguishable from the real thing.

The Chinese Room: 

The Chinese Room Chapter 6 in Copeland (1993): The curious case of the Chinese Room. Clearer account: pgs 292-297 in Sharples, Hogg, Hutchinson, Torrance and Young (1989) ‘Computers and Thought’ MIT Press: Bradford Books. Original source: Minds, Brains and Programs: John Searle (1980) Important philosophical critic of Artificial Intelligence. See also more recent book: Searle, J.R. (1997) The Mystery of Consciousness. Granta Books, London

Slide59: 

Searle is an opponent of strong AI, and the Chinese room is meant to show what strong AI is, and why it is wrong. NB. It isn’t clear that there are ANY people who really believe strong AI… Searle makes the assumption that a computer program based on the following principle WILL pass the Turing Test.

Slide60: 

Symbols mean nothing to operator. Instructions correspond to program which simulates linguistic ability and understanding of native speaker of Chinese. Sets of symbols passed in and out correspond to sentences of meaningful dialogue.

Slide61: 

More than this: Chinese Room program is (perhaps!) able to pass the Turing Test with flying colours! According to Searle, behaviour of operator is like that of computer running a program. What point do you think Searle is trying to make with this example?

Slide62: 

Operator does not understand Chinese - only understands instructions for manipulating symbols. Behaviour of operator is like behaviour of computer running same program. Computer running program does not understand any more than the operator does. Operator only needs syntax, not semantics.

Slide63: 

Syntax knowledge of formal properties of symbols (how they can be combined). Semantics relating symbols to real world. Mastery of syntax: mastery of set of rules for performing symbol manipulations. Mastery of semantics: to have understanding of what those symbols mean (this is the hard bit!!)

Turing’s own objections:: 

Turing considered, and dismissed, some possible objections to the idea that computers can think. Some are easier to refute than others. Objections considered by Turing: The theological objection The ‘heads in the sand’ objection The mathematical objection The argument from consciousness Arguments from various disabilities Lady Lovelace’s objection Argument from continuity in the nervous system The argument from informality of behaviour The argument from extra-sensory perception Turing’s own objections:

The theological objection: 

The theological objection ‘…Thinking is a function of man’s immortal soul. God has given an immortal soul to every man and woman, but not to any other animal or to machines. Hence no animal or machine can think…’ BUT:Why not believe that God could give a soul to a machine if He had wished to? back

Heads in the sand objection: 

i.e. The consequence of machines thinking would be too dreadful. Let us hope and believe that they cannot do so. This is related to the theological argument; idea that humans are superior to the rest of creation, and must stay so……... ‘.. Those who believe in ..(this and the previous objection).. would probably not be interested in any criteria [for deciding if machines could think].’ back Heads in the sand objection

The mathematical objection: 

The mathematical objection Results in mathematical logic which can be used to show that there are limitations to the powers of discrete-state machines. Goedel’s Theorem For any formal axiomatic system F powerful enough to do arithmetic Goedel sentence: G(F) If F is consistent, then G(F) is true

Slide68: 

The Halting Problem: will the execution of a program P eventually halt or will it run for ever? Turing (1936) proved that for any algorithm H that purports to solve halting problems there will always be a program Pi such that H will not be able to answer the halting question correctly Hence certain questions cannot be answered correctly by any formal system (cf. Goedel) But, a Liar’s Paradox sentence cannot be consistently asserted by an agent human or otherwise: Turing cannot consistently assert that this sentence is true. And also, similar limitations may also apply to the human intellect. Lots of people have written on this (look at Hofstader’s Goedel, Escher, Bach) back

The argument from consciousness: 

The argument from consciousness ‘…This argument is very well expressed in Professor Jefferson’s Lister Oration for 1949, from which I quote. “Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain – that is not only write it but know that it had written it. No mechanism could feel (and not merely artificially signal, an easy contrivance) pleasure at its successes, grief when its valves fuse, be warmed by flattery, be made miserable by its mistakes, be charmed by sex, be angry or depressed when it cannot get what it wants”..’ From Turing’s paper

But…...: 

But…... The only way one could be sure that a machine thinks is to be that machine and feel oneself thinking. - similarly, only way to be sure someone else thinks, is to be that person. How do we know that anyone is conscious? (=solipsism). Instead, we assume that others can think and are conscious----it is a polite convention. Similarly, could assume that machine which passes Turing test is so too. back

Arguments from various disabilities: 

Arguments from various disabilities ie ‘I grant that you can make machines to all the things you have mentioned but you will never be able to make one do X’. eg be kind, resourceful, beautiful, friendly, have initiative, have a sense of humour, tell right from wrong, make mistakes, fall in love, enjoy strawberries and cream, make someone fall in love with it, learn from experience, use words properly, be the subject of its own thought, have as much diversity of behaviour as a man, do something really new. These criticisms often disguised forms of argument from consciousness. back

Lady Lovelace’s objection:: 

Lady Lovelace’s objection: (memoir from Lady Lovelace about Babbage’s Analytical Engine - 1842) Babbage (1792-1871) and Analytical Engine: general purpose calculator. Entirely mechanical. Contraption never built – engineering not up to it and no electricity! ‘..The Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform..’ Objection: A computer cannot be creative, it cannot originate anything, only carry out what was given to it by the programmer.

But…...: 

But…... computers can surprise their programmers. – I.e. by producing answers that were not expected. Original data may have been given to computer, but may then be able to work out its consequences and implications (cf. level of chess programs and their programmers). back

Argument from continuity in the nervous system: 

Argument from continuity in the nervous system Nervous system is continuous: the digital computer is discrete state machine. I.e. in the nervous system a small error in the information about the size of a nervous impulse impinging on a neuron may make a large difference to the size of the outgoing impulse.

Slide75: 

Discrete state machines: move by definite transitions from one state to another. For example, consider the ‘convenient fiction’ that switches are either definitely on, or definitely off. However, discrete state machine can still give answers that are indistinguishable from a continuous machine. back

Turing Vs Searle: 

Turing Vs Searle Turing: ‘Can machines think?’ was better seen as a practical/empirical question, so as to avoid the philosophy (it didn’t work!). Empirical – tested by experiment Searle - ‘can a machine think’ is not an empirical question. Something following a program could never think.

Slide77: 

Strong AI: Machine can literally be said to understand the responses it makes. Searle’s argument is that like the operator in the Chinese Room, A Chatbot does not understand anything it responds--is it true in principle, as Searle wants?

Slide78: 

Questions: is Searle’s argument convincing? Does it capture some of your doubts about computer programs?

Slide79: 

Suppose for a moment Turing had believed in Strong AI. He might have argued: a computer succeeding in the imitation game will have same mental states that would have been attributed to human. Eg. understanding the words of the language been used to communicate. But, says Searle. the operator cannot understand Chinese.

Slide80: 

Treat the Chinese Room system as a black box and ask it (in Chinese) if it understands Chinese - “Of course I do” Ask operator (if you can reach them!) if he/she understands Chinese - “search me, its just a bunch of meaningless squiggles”.

Other objections: 

Copeland (1993) [see ‘Artificial Intelligence: a philosophical introduction’] 4 further objections to Turing Test. The first three of these he dismisses, and the fourth he incorporates into a modified version of the Turing Test. Other objections

1. Too conservative: Chimpanzee objection: 

1. Too conservative: Chimpanzee objection Chimpanzees, dolphins, dogs, and pre-linguistic infants all can think (?) but could not pass Turing Test. But this only means that Turing Test cannot be a litmus test (red = acid, not red = non acidic). - nothing definite follows if computer/animal/baby fails the test. i.e. negative outcome does not mean computer cannot think. (In philosophical terms: TT gives a sufficient not a necessary condition of thought)

2. Too easy: Sense organs objection: 

2. Too easy: Sense organs objection Turing Test focuses only on verbal responses: the computer is able to use words without really knowing their meanings. (like taking a driving test that consists only of answering questions).

Slide84: 

Possible answer: Equip it with sense organs, and you can then test knowledge of world And there are huge numbers of words which can be investigated without having sense organs. Eg ‘teasing’. Think about getting to know people through web chat--can you?

3. Simulation objection: 

3. Simulation objection The simulation of X is never an X. In the original Victorian parlour game, if a man misleads the interrogator into thinking he is a woman, it doesn’t mean he IS a woman.

Slide86: 

But there are two kinds of simulation: a) simulation1 which lacks essential features of what is being simulated. Eg. in simulated death the person is not dead. b) simulation2, exactly like what is being simulated, but not produced in standard way (e.g. aircraft in wind tunnels). One could claim that a program that passed the Turing Test was a simulation2

4. Black box objection.: 

4. Black box objection. This objection is presented by Copeland (1993), and not rejected by him. A black box is a device whose inner workings are unknown. The Turing Test: relies only on outward behaviour. But one might look inside computer, to see how it ‘thinks’. i.e. might have program which just has list of sentence to be output in response to sentences it encounters. All possibilities covered, and computer not thinking at all, just matching sentences.

The First ChatBots: 

The First ChatBots Eliza and Parry

ELIZA (1964): 

ELIZA (1964) Joseph Weizenbaum’s program, a parody of a Rogerian therapist, has been available for decades Now distributed as part of EMACS text editor Plenty of web links It had a major impact on computer games, like Dungeon (1975)

Slide90: 

Men are all alike. IN WHAT WAY? They’re always bugging us about something or other. CAN YOU THINK OF A SPECIFIC EXAMPLE? Well, my boyfriend made me come here. YOUR BOYFRIEND MADE YOU COME HERE? He says I’m depressed much of the time. I AM SORRY TO HERE THT YOU ARE DEPRESSED. It’s true that I am unhappy. DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE UNHAPPY? I need some help, that much seems certain. WHAT WOULD IT MEAN TO YOU IF YOU GOT SOME HELP? Perhaps I could learn to get along with my mother. TELL ME MORE ABOUT YOUR FAMILY. My mother takes care of me.

Slide91: 

WHO ELSE IN YOUR FAMILY TAKES CARE OF YOU? My father. YOUR FATHER You are like my father in some ways. WHAT RESEMBLANCE DO YOU SEE? You are not very aggressive, but I think you don’t want me to notice that. WHAT MAKES YOU THINK I AM NOT VERY AGGRESSIVE? You don’t argue with me. WHY DO YOU THINK I DON’T ARGUE WITH YOU? You are afraid of me. DOES IT PLEASE YOU TO THINK I AM AFRAID OF YOU? My father is afraid of everybody. WHAT ELSE COMES TO MIND WHEN YOU THINK OF YOUR FATHER Bullies DOES THAT HAVE ANYTHING TO DO WITH THE FACT THAT YOUR BOYFRIEND MAKE YOU COME HERE?

Is ELIZA impressive?: 

Is ELIZA impressive? When we look at how Eliza works, the conversation is even less impressive. If pattern matches then give stored response Otherwise give general purpose response (X me Y) ---> (X you Y) (I remember X) ---> (Why do you remember X just now?) (My {family-member} is Y) ---> (Who else in your family is Y?) (X {family-member} Y) ---> (Tell me more about your family)

Slide93: 

Eliza lacks Means of recognizing grammatical structure of language Means of converting users query into a representation of the underlying knowledge from which inferences can be drawn.

Slide94: 

BUT We don’t have any INDEPENDENT evidence that people do assign grammatical structures to what they hear in dialogue, nor that they use a great deal of background knowledge when carrying on a superficial conversation…

PARRY (1971): 

PARRY (1971) Written by Kenneth Colby, a psychiatrist at Stanford in 1972. PARRY had far more to say than ELIZA and wanted to tell you things It modelled a psychiatrically disordered person (PARanoid) and hence it was easy to excuse when it didn’t make sense. It remembered what you had said and got ‘annoyed’ and ‘frightened’ Colby ran a series of TT’s with psychiatrists – PARRY passed

Conversation with Parry: 

Conversation with Parry Why are you in the hospital? I SHOULDN’T BE HERE. Who brought you here? THE POLICE. What trouble did you have with the police? COPS DON’T DO THEIR JOB. What do they do instead? COPS ARREST THE WRONG PEOPLE. They arrest the right ones sometimes.

Slide97: 

THE LAW SHOULD GET THOSE ITALIAN CROOKS. Are you Italian? MY BACKGROUND IS BRITISH BUT I WAS BORN IN THIS COUNTRY. Do you know any Italian crooks? I TRY TO AVOID THE UNDERWORLD. Are you afraid of the underworld? THE MAFIA ARE THE WORST GANGSTERS. Are you connected with the Mafia? YOU KNOW, THEY KNOW ME.

Slide98: 

How do they know you? THEY SPY ON ME. Why do they? THEY WANT TO GET FALSE EVIDENCE TO FRAME ME. What do they have against you? THEY ARE OUT TO GET ME. What for? I HAVE INCRIMINATING EVIDENCE AGAINST THE MAFIA.

Slide99: 

Are you sure of all this? YOU DON’T BELIEVE ME? I am doing my best. …silence CAN’T YOU THINK OF A DECENT QUESTION TO ASK? You seem hostile to me. SOME OF YOUR REMARKS ANNOY ME. I am sorry. FORGET THE EXCUSES, YOU GUYS ARE ALL ALIKE.

Colby’s PARRY : 

Colby’s PARRY Perhaps best ever performance, many users, robust, but not a normal subject (I.e. a paranoid) primitive individual models, some control of dialogue process; but it had lots to say! not syntax but fast pattern matching much better than ELIZA

Parry’s Internal Graph: 

Parry’s Internal Graph Gambling Money Police Italians

Parry & Eliza: 

Parry & Eliza parry: People get on my nerves sometimes. eliza: I am not sure I understand you fully. parry: You should pay more attention. eliza: Suppose you should pay more attention? parry: You're entitled to your opinion. eliza: What makes you think I'm entitled to my opinion? parry: It bothers me just to be around people. eliza: Please go on.

Slide103: 

parry: You harp on the same questions too much. eliza: What does that similarity suggest to you? parry: I don't get you. eliza: Do you wish that you did get it? parry: It's about time you showed some interest in my feelings. eliza: Oh, I showed some interest in your feelings?

Should the Turing Test be abandoned?: 

Should the Turing Test be abandoned? Reason 1: Unitary notion of ‘intelligence’ too simplistic. Too simplistic to think that it is useful to assess whether computers possess ‘intelligence’, or the ability to think. Better to break down this question into smaller questions. similar to idea that unitary measure of intelligence (ie intelligence as measured by IQ tests) is not very useful better to have tests that reveal the relative strengths and weaknesses of individuals. Could assess computers in terms of more specific abilities; ability of robot to navigate across a room ability of computer to perform logical reasoning metaknowledge (knowledge of own limitations).

Slide105: 

Reason 2: Too anthropocenctric. Too anthropocentric to insist that program should work in same way as humans. Dogs are capable of cognition, but would not pass Turing Test. Producing machine with cognitive and communicative abilities of a dog would be (another) challenge for AI. But how can we NOT be anthropocentric about intelligence? We are the only really intelligent things we know, and language is closer to our intelligence than any other function we have…? 

…behaviour is all we have…: 

…behaviour is all we have… Until opening heads actually works Increasingly complex programs means that looking inside machines doesn’t tell you why they are behaving the way they are. Those who don’t think the TT effective must show why machines are in a different position from our fellow humans.

back to PARRY: 

back to PARRY PARRY was not designed to show understanding, but was often thought to do so. We know it worked with a very simple but large mechanism: Why are you in the hospital? I SHOULDN’T BE HERE. Who brought you here? THE POLICE. What trouble did you have with the police? COPS DON’T DO THEIR JOB.

Responses to Searle:: 

Responses to Searle: Insist that the operator can in fact understand Chinese Like case in which person plays chess who does not know rules of chess but is operating under post-hypnotic suggestion.

Slide109: 

Systems Response concede that the operator does not understand Chinese, but that system as a whole, of which operator is a part, DOES understand Chinese. Premiss: Bill the cleaner has never sold pyjamas to Korea. Therefore the company for which Bill works has never sold pyjamas to Korea.

Searle’s rebuttal: 

if symbol operator doesn’t understand Chinese, why should you be able to say that symbol operator plus bits of paper plus room understands Chinese. Therefore no amount of symbol manipulation on operators part will enable the wider system of which operator is a component to understand the Chinese input. Searle claims system as a whole behaves as though it understands Chinese. But that doesn’t mean that it does. Searle’s rebuttal

The Internalised Case: 

Suppose the operator learns up all these rules and table and can do the trick in Chinese. On this version, the Chinese Room has nothing in but the operator. Can one still say the operator understands nothing of Chinese? Consider: a man appears to speak French fluently but say, no he doesn’t really, he’s just learned up a phrase book. He’s joking, isn’t he? The Internalised Case

Slide112: 

You cannot really contrast a person with rules-known-to-the person In Noam Chomsky’s view language behaviour in humans IS rule based (and he can determine what the rules are!)

Recent restatement of Chinese Room Argument: 

Recent restatement of Chinese Room Argument From Searle (1997) The Mystery of Consciousness 1. Programs are entirely syntactical 2. Minds have a semantics 3. Syntax is not the same as, nor by itself sufficient for, semantics Therefore programs are not minds. QED

1. Programs are entirely syntactical: 

states that a program written down consists entirely of rules concerning syntactical entities, that is rules for manipulating symbols. Physics of implementing medium (ie computer) is irrelevant to computation. 1. Programs are entirely syntactical

2. Minds have a semantics: 

2. Minds have a semantics says what we know about human thinking. When we think in words or other symbols we have to know what those words mean - a mind has more than uninterpreted formal symbols running through it, it has mental contents or semantic contents.

3. Syntax is not the same as, nor by itself sufficient for, semantics: 

states the general principle that Chinese Room thought experiment illustrates. Merely manipulating formal symbols does not guarantee presence of semantic contents. 3. Syntax is not the same as, nor by itself sufficient for, semantics

Summary: 

Summary ‘..It does not matter how well the system can imitate the behaviour of someone who really does understand, nor how complex the symbol manipulations are; you can not milk semantics out of syntactical processes alone..’ (Searle, 1997)

More on Semantics…: 

Searle says this shows the need for semantics but semantics means two things at different times: Access to objects via FORMAL objects (more symbols) as in logic and the formal semantics of programs. Access to objects via physical contact and manipulation--robot arms or prostheses (or what children do from a very early age). More on Semantics…

Semantics Issues: 

Programs have access only to syntax. If Searle is offered a formal semantics “that’s just more symbols.” If offered access to objects via a robot prothesis from inside the box “just more programs or it won’t have reliable reference like us.” Semantics Issues

Slide120: 

Is there any solution to the issues raised by Searle’s Chinese Room? Are there any ways of giving the symbols real meaning? Does it matter?????????????