CMSC 471

History of AI
Chronology of AI, R&N Chapter 26
Adapted from slides by Tim Finin and Marie desJardins.
1

Todays class
History of AI
Key people Significant events

Future of AI
Where are we going

Philosophy of AI
Can we build intelligent machines?
If we do, how will we know theyre intelligent?

Should we build intelligent machines?

If we do, how should we treat them and how will they treat us?

Key people (AI prehistory)

George Boole invented propositional logic (1847) Karel Capek coined the term robot (1921) Isaac Asimov wrote many sf books and essays (I, Robot (1950) introduced the Laws of Robotics if you havent read it, you should!) John von Neumann: minimax (1928), computer architecture (1945) Alan Turing: universal machine (1937), Turing test (1950) Norbert Wiener founded the field of cybernetics (1940s) Marvin Minsky: neural nets (1951), AI founder, blocks world, Society of Mind John McCarthy invented Lisp (1958) and coined the term AI (1957) Allen Newell, Herbert Simon: GPS (1957), AI founders Noam Chomsky: analytical approach to language (1950s)
3

Key people (early AI history)

Hubert and Stuart Dreyfus: anti-AI specialists Ed Feigenbaum: DENDRAL (first expert system, 1960s) Terry Winograd: SHRDLU (blocks world, 1960s) Roger Schank: conceptual dependency graphs, scripts (1970s) Shakey: mobile robot (SRI, 1969) Doug Lenat: AM, EURISKO (math discovery, 1970s) Ed Shortliffe, Bruce Buchanan: MYCIN (uncertainty factors, 1970s)

Key events: Genesis of AI

Turing test, proposed in 1950 and debated ever since Neural networks, 1940s and 1950s, among the earliest theories of how we might reproduce intelligence Logic Theorist and GPS, 1950s, early symbolic AI Dartmouth University summer conference, 1956, established AI as a discipline Early years: focus on search, learning, knowledge representation Development of Lisp, late 1950s

Key events: Adolescence of AI

The movie 2001: A Space Odyssey (1968) brought AI to the publics attention Early expert systems: DENDRAL, Meta-DENDRAL, MYCIN Arthur Samuelss checkers player, Doug Lenats AM and EURISKO systems, and Werboss and Rumelharts backpropagation algorithm held out hope for the ability of AI systems to learn Hype surrounding expert systems led to an inevitable decline in interest in the mid to late 1980s, when it was realized they couldnt do everything Hype surrounding neural networks in the late 1980s led to similar disappointment in the 1990s Roger Schanks conceptual dependency theory and Doug Lenats Cyc started to address problems of common-sense reasoning and representation Hans Berliners heuristic search player defeated the world backgammon champion in 1979
6

Key events: AI adulthood (barely)

Many commercial expert systems introduced, especially in the 1970s and 1980s Fuzzy logic and neural networks used in controllers, especially in Japan and Europe Recent developments and areas of great interest include:
Bayesian reasoning and Bayes nets Ontologies, knowledge reuse, and knowledge acquisition Mixed-initiative systems that combine the best of human and computer reasoning Multi-agent systems, Internet economies, intelligent agents Autonomous systems for space exploration, search and rescue, hazardous environments
7

What do AI people do?

Subject headings from IJCAI-01 conference proceedings:
Knowledge Representation and Reasoning Search, Satisfiability, and Constraint Satisfaction Cognitive Modeling Planning Games Diagnosis Logic Programming and Theorem Proving Uncertainty and Probabilistic Reasoning Neural Networks and Genetic Algorithms Machine Learning and Data Mining Case-Based Reasoning Multi-Agent Systems Natural Language Processing and Information Retrieval Robotics and Perception Web Applications

Are we there yet?

Great strides have been made in knowledge representation and decision making Many successful applications have been deployed to (help) solve specific problems Key open areas remain:
Incorporating uncertain reasoning Real-time deliberation and action Perception (including language) and action (including speech) Lifelong learning / knowledge acquisition Common-sense knowledge Methodologies for evaluating intelligent systems

Philosophy of AI
Alan M. Turing, Computing Machinery and Intelligence John R. Searle, Minds, Brains, and Programs

10

Philosophical debates
What is AI, really?
What does an intelligent system look like? Does an AI needand can it haveemotions, consciousness, empathy, love?

Can we ever achieve AI, even in principle? How will we know if weve done it? If we can do it, should we?

11

Turing test
Basic test:
Interrogator in one room, human in another, system in a third Interrogator asks questions; human and system answer Interrogator tries to guess which is which If the system wins, its passed the Turing Test

The system doesnt have to tell the truth (obviously)

12

Turing test objections

Objections are basically of two forms:
No computer will ever be able to pass this test Even if a computer passed this test, it wouldnt be intelligent

Chinese Room argument (Searle, 1980), responses, and counterresponses

Robot reply Systems reply

13

Machines cant think

Theological objections Its simply not possible, thats all Arguments from incompleteness theorems
But people arent complete, are they?

Machines cant be conscious or feel emotions

Reductionism doesnt really answer the question: why cant machines be conscious or feel emotions??

Machines dont have Human Quality X Machines just do what we tell them to do
Maybe people just do what their neurons tell them to do

Machines are digital; people are analog

14

The Turing test isnt meaningful

Maybe so, but

If we dont use the Turing test, what measure should we use?

Very much an open question

15

Ethical concerns: Robot behavior

How do we want our intelligent systems to behave? How can we ensure they do so? Asimovs Three Laws of Robotics:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm. 2. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. 3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

16

Ethical concerns: Human behavior

Is it morally justified to create intelligent systems with these constraints?
As a secondary question, would it be possible to do so?

Should intelligent systems have free will? Can we prevent them from having free will?? Will intelligent systems have consciousness? (Strong AI)
If they do, will it drive them insane to be constrained by artificial ethics placed on them by humans?

If intelligent systems develop their own ethics and morality, will we like what they come up with?

17