|
|
|
| Current Issue |
|
|
| Resources |
|
|
|
The Artificiality of Artificial Intelligence
| Article
# : |
10463 |
|
|
Section : |
NATURAL SCIENCE
|
| Issue
Date : |
2 / 1993 |
2,023 Words |
| Author
: |
Greg Smith and David Kruse
Greg Smith is a prelaw student in David Kruse's physics class
at Colby College. David Kruse teaches computer science and
physical science classes at Colby College in Colby, Kansas. |
Marvin Minsky, a leading pioneer in the field of artificial intelligence (AI), not so long ago made the claim that within a few short years computers would far surpass their human creators in intelligence. Today, as computer scientists struggle to even agree on a definition of what artificial intelligence is, it is painfully obvious that endowing a computer with intelligence is a feat that has eluded us so far.
The answers to ancient questions like "what is knowledge?" and "What is intelligence?" so intrinsic in defining where the course of AI research will lead, are probably keys to understanding how our own minds work and, consequently, how the "mind" of a computer could work. If we really understood the underlying mechanisms of intelligence and could master the intricacies of human thought and cognition, modeling a computer after ourselves would not be such an august task.
Although many theories have been proffered to define what AI is, none has gained a majority consensus among AI practitioners. The problem is exacerbated by the extreme difference of thought between AI researchers. Modern phenomenologist and Berkeley Prof. Hubert Dreyfus opines that the level of machine intelligence is a fixed value and that today's limits are boundaries past which AI will never journey. He is countered by others such as Minsky who feel that building a machine with true intelligence is possible, if only because we ourselves are intelligent.
Perhaps a good definition of AI, and possibly a goal to strive for in creating artificially intelligent computers, is an entity that expresses cognition rather than mere computation, and semantics rather than superficial syntax.
Slippability
While this definition seems logical and may be partially correct, porting pure intelligence over into the realm of artificial intelligence needs something else entirely. Douglas Hofstadter, professor of cognitive science at Indiana University, calls this extra ingredient "slippability," meaning the nonrigidness of thought that sometimes accompanies the human brain. This is analogous to heuristics and the ability to make a decision by using a more "humanlike" method than just applying pure logic to the problem at hand. It might also imply the fact that an intelligent being should be able to make a mistake, just as we imperfect humans intermittently do.
At this stage in the AI game,
...
Read Full Article
Look for this article in Ask.com
|
|
