Lecture 27 Social Implications of Artificial Intelligence Social & Legal Issues

advertisement
Lecture 27
4/19/04 22:09
Lecture 27
Social & Legal Issues
(S&G, ch. 14)
4/19/04
CS 100 - Lecture 27
1
Social Implications of
Artificial Intelligence
4/19/04
CS 100 - Lecture 27
2
1
Lecture 27
4/19/04 22:09
Kismet (Brooks’ Lab, MIT)
• Responds to a face
with a happy
expression
• Responds to rapidly
moving face with
disgusted expression
4/19/04
CS 100 - Lecture 27
3
Kismet (Brooks’ Lab, MIT)
• Example of three-way
conversational
interaction
4/19/04
CS 100 - Lecture 27
4
2
Lecture 27
4/19/04 22:09
Social Implications of AI
• Elimination of jobs
• Replacing human judgment by machine
judgment
• Altering our understanding of ourselves
4/19/04
CS 100 - Lecture 27
5
Consider Computers that are
More Intelligent than Humans
• Suppose it’s impossible
– establishing this would require some kind of scientific
proof that a computer cannot emulate a brain, or that a
brain is not sufficient for intelligence
– in either case, a major scientific breakthrough
• Suppose it’s possible
– could displace us in all intellectual activities
– if we had nothing to do, what would it mean to be
human?
– would/should humans be obsolete?
4/19/04
CS 100 - Lecture 27
6
3
Lecture 27
4/19/04 22:09
Social & Legal Issues
4/19/04
CS 100 - Lecture 27
7
Promethean Fire
• “All technology is non-neutral”
• Computer technology has:
– social benefits
– social costs
• Often, issues are not peculiar to computers,
but are amplified by computer technology
• Sometimes problems arise from using
computers as they are intended to be used
4/19/04
CS 100 - Lecture 27
8
4
Lecture 27
4/19/04 22:09
Privacy
•
•
•
•
Government databases (federal & state)
Computer matching
Computer profiling
Open & closed files
4/19/04
CS 100 - Lecture 27
9
Uses of Personal Data
• Marketing
• Decision making
• Other uses:
– secondary use
– invisible information gathering
– information from children
• “Information underground”
4/19/04
CS 100 - Lecture 27
10
5
Lecture 27
4/19/04 22:09
Protecting Privacy
•
•
•
•
•
Access control
Audit trails
Encryption
Ethical use policies
Informed consent
4/19/04
•
•
•
•
Regulation
Ownership of data
Contracts
Markets, options &
consumer pressure
CS 100 - Lecture 27
11
Computer Error
•
•
•
•
Overdependence on computers
Major failures
Lesser failures
Limitations of computer simulation
4/19/04
CS 100 - Lecture 27
12
6
Lecture 27
4/19/04 22:09
Simulation & Models
• Recall our discussion of models:
– a model is intended to address a certain class of
questions or issues
– models make simplifying assumptions
• Recall the omission of aerodynamic
stability from bridge models before the
Tacoma Narrows Bridge disaster
• Nevertheless, computer models &
simulations are essential
4/19/04
CS 100 - Lecture 27
13
Evaluating Models
• How well is the underlying science
understood?
• What sorts of simplifying assumptions are
made?
• How well does the model agree with
experimental results?
4/19/04
CS 100 - Lecture 27
14
7
Lecture 27
4/19/04 22:09
Computer Crime
• Often just a new or “better” way to commit
traditional crimes
• Ampliative aspects of computer technology
– may allow committing crimes on a larger scale
• Reductive aspects of computer technology
– may make it harder to detect crimes, identify
criminals, etc.
4/19/04
CS 100 - Lecture 27
15
Some Kinds of Computer Crime
• Fraud, embezzlement & theft
• Stealing (data)
• “Hacking” (system cracking)
– individual systems
– the information infrastructure
– benign hacking?
• Viruses & worms: reproduce selves
– virus: hides in another program
– worm: an independent program
4/19/04
CS 100 - Lecture 27
16
8
Lecture 27
4/19/04 22:09
Constitutional & Civil Liberties
Issues
• 1st Amendment protects freedom of speech
& press
• 4th Amendment protects against
unreasonable search & seizure
• Do they apply to electronic & other new
media?
• So far, the courts generally have held them
to apply, but there is much uncertainty
4/19/04
CS 100 - Lecture 27
17
Communication Media
• Is the operator of a communication medium
responsible for the information it carries?
– analogy: telephone companies & post office
– analogy: a privately-owned meeting room
• Can the operator restrict this information?
– analogy: telephone companies & post office
– analogy: private presses
4/19/04
CS 100 - Lecture 27
18
9
Lecture 27
4/19/04 22:09
Encryption & Wiretapping
• Encryption can be used to ensure privacy &
authenticity
– may be used for legal/illegal purposes
• May be subject to export restrictions
• What provisions should be made for
government interception & recording of
communications?
• What limits should be imposed on the
government
4/19/04
CS 100 - Lecture 27
19
10
Download