14-HumanMachines.ppt

Department of Computer ScienceDepartment of Computer Science

City College of New YorkCity College of New York

Spring 2006Spring 2006

Copyright Copyright © 2006 by Abbe Mowshowitz© 2006 by Abbe Mowshowitz

CSc 375CSc 375SOCIAL ISSUES IN COMPUTINGSOCIAL ISSUES IN COMPUTING

TOPIC 14. HUMAN TOPIC 14. HUMAN MACHINESMACHINES

A. Intellectual Challenge to Human Identity

1. “Man is nothing but a meat machine.”

- Marvin Minsky

HUMAN MACHINESHUMAN MACHINES

A. Intellectual Challenge to Human Identity

2. Strong form of Church’s Thesis:

Any effective procedure can be programmed on a computer.

ANDEverything humans can do can be expressed as effective procedures.

HUMAN MACHINESHUMAN MACHINES

A. Intellectual Challenge to Human Identity

3. Turing Test (Alan Turing. “Computing machinery and intelligence, 1950.): reformulation of the question “can machines think?”

HUMAN MACHINESHUMAN MACHINES

A.3. Turing Test Based on the “Imitation Game” in whichHuman interrogator converses by means of a tele-

typewriter with two unidentified respondents

- one a machine

- one a person

Object: distinguish between person and machine

HUMAN MACHINESHUMAN MACHINES

A.3. Turing Test

If in repeated trials with different humansubjects, the interrogator cannot distinguishwith better than 50% accuracy, the machineis said to simulate human intelligence.

Japanese scientists have unveiled a "female" android called Repliee Q1.

Source: BBCNews, July 28, 2005

“Professor Hiroshi Ishiguro of Osaka University saysone day robots could fool us into believing theyare human.”

HUMAN MACHINESHUMAN MACHINES

A. Intellectual Challenge to Human Identity

4. Information processing model (Simon and Newell, 1964)

a. Science of information processing (IP) can be independent of particular IP mechanisms

b. Thinking can be explained in IP terms

c. IP theories can be formulated in programming languages and simulated on computer

HUMAN MACHINESHUMAN MACHINES

B. Practical Challenge Simon’s continuum of behavior:

Capabilities of computer applications have been moving steadily toward the non-programmed end since the 1960s

PROGRAMMED NON-PROGRAMMED

HUMAN MACHINESHUMAN MACHINES

C. Historical Perspective– Earth pushed from center stage by the

heliocentric theory of Copernicus– Humans linked to apes by Darwin– Denied conscious control by Freud– Now AI would turn us into machines

without free will

HUMAN MACHINESHUMAN MACHINES

D. Challenge Posed by AI1. Until 1980s AI was an obscure academic discipline2. Support for research came mostly from government (DoD in the US)• Intelligent command and control• Guidance for automatic weapons• Spying, reconnaissance

HUMAN MACHINESHUMAN MACHINES

D.2. Support

spying/reconnaissance calls for:• Natural language processing• Speech recognition• Image processing

Examples: unmanned, armored tank; automated co-pilot

HUMAN MACHINESHUMAN MACHINES

D. Challenge Posed by AI4. AI in the spotlight• Japan Fifth Generation Computer Project

– Announced 1980 by MITI– Projected as 10 year $500 m. program– Proposed to develop revolutionary computer

systems incorporating AI concepts: problem solving functions, intelligent interfaces, inference & knowledge-based functions

HUMAN MACHINESHUMAN MACHINES

D.4. AI in the spotlight• U.S. answer to Fifth Generation Project

– DARPA’s Strategic Computing Program– $600 m. funding– Envisioned developing AI systems and

advanced computer technology

HUMAN MACHINESHUMAN MACHINES

D. Commercial Developments in AI

1. Emergence of start-up firms marketing expert systems with• Knowledge base containing rules of thumb• Inference engine that makes decisions based

on rules in a specialized domain

Major challenge: capturing knowledge of human expert and embedding in system (knowledge engineering)

HUMAN MACHINESHUMAN MACHINES

D. Commercial Developments in AI2. Expert system applications are now widespread (e.g., authorizer’s assistant)

3. Motivation: reduce labor costs, provide consistent performance

HUMAN MACHINESHUMAN MACHINES

D. Differences in Cognitive Capabilities of Humans and Machines?1. Affirmative answer calls for rejection of Church’s Thesis (e.g., Dreyfus’ argument based on need for a body)2. What cannot be done today may be realized tomorrow?

HUMAN MACHINESHUMAN MACHINES

E. Ethical Implications of AI

1. Robots and ‘ethical brakes’

(e.g., Asimov’s three laws of robotics)

2. Are there areas of application that should be off limits to machines?• Judgment in criminal cases?• Psychotherapy?

HUMAN MACHINESHUMAN MACHINES

F. Literary Perspectives

1. Background• Early Post WW II period• Stimulated by the first computers• Themes echo those in science fiction

and futuristic literature

HUMAN MACHINESHUMAN MACHINES

F.2. Central Figures• Robots

– Capek’s R.U.R. (Rossum’s Universal Robots), 1923

– Czech word “robota” means work– Robots limited in action by design– Usually obedient servants

HUMAN MACHINESHUMAN MACHINES

F.2. Central Figures• Androids (or humanoids): living beings

not created by human birth• Computers: based loosely on

contemporary machines• Hybrids (cyborgs): mixed organic and

computer components

Distinctions hard to maintain in practice.

HUMAN MACHINESHUMAN MACHINES

F.3. Major Themes• Dehumanization

– Over-reliance on machines– Disregard of human needs– Excessive standardization and inflexibility

• Identity crisis– Superfluity (nothing to do)– Myth of regeneration

HUMAN MACHINESHUMAN MACHINES F.3. Major Themes

• Persistence of human impulses (especially in negative utopias)

• Unanticipated consequences (Pandora’s Box)

• Knowledge and power (cosmic mind, computer as God)

– Forbidden knowledge and the Word of God– Sorcerer’s apprentice, Golem– Computers as adjunct to social power

HUMAN MACHINESHUMAN MACHINES

F.3. Major Themes

• Partnership – Human-robot interaction– Synthesis of man and machine into higher

entity

HUMAN MACHINESHUMAN MACHINES

F. Literary Perspectives

4. Views and attitudes• Ambivalence towards technology• Machine take-over of human functions• Anxiety over role of humans in society• Fear of diminished human worth and

dignity