Transcript Killing HAL

Alexandr Gamezo
 Talos of Crete
 Galatea of Pygmalion
 Frankenstein
 Field began in 1956 at Dartmouth College
 “Fathers” of the field were John McCarthy, Marvin
Minsky, Allen Newell, and Herbert Simon
 The field saw waves of interest.
 First peak was in the 1960s. US DoD heavily funded
research
 1980s saw a peak due to development of Expert Systems.
 In the 1990’s and early 21st century, AI was integrated
into many fields throughout the technology industry
 Logistics
 Data Mining
 Medical Diagnosis
 Aided by the huge leap in computational power and
the emphasis of the field on solving specific problems
 No universally accepted definition
 Some generally accepted requirements are:
 Deduction, reasoning
 Planning
 Learning
 Perception
 Social Intelligence
 Creativity
 Computers have many of the traits that are required
for “intelligence”
 Computers can reproduce any human logical reasoning
 Computers can plan paths to solutions given enough
information
 Computers can be made to learn and alter their own
programming to incorporate that
 Kismet is a robot at MIT that has been made to show
and understand emotion and body language
 Furbies
 Will we recognize the intelligence of a synthetic life
when we see/develop it?
 Turing Test is only current test of Artificial Intelligence
 Compare “knowledge” with a person in another
country
 Ray Kurzweil estimates that by 2029 desktop
computers will have same power as human brains and
by 2045 computers will improve themselves faster than
we ever could
 Developers may not see their software as sentient
beings
 Fully self-aware beings could be “deleted” simply due
to not being convincingly “human”
 Just how intelligent is intelligent?
 How will we detect intelligence in something that





could be vastly different from our own intelligence?
What position will synthetic intelligence have in our
society?
Will they be our new slave labor?
Will they be our allies and assistants?
How and will they be made docile toward people?
Research into this should be at least parallel with
actual development of AI.
 Technology is close to fulfilling the dream of an actual
synthetic intelligence
 We have no real method of testing that intelligence
 We must find answers to some very basic ethical
questions before we continue to play God