The Computer Revolution/Artificial Intelligence/Weak VS Strong AI

From Wikibooks, open books for an open world
< The Computer Revolution
Jump to: navigation, search

Weak Artificial Intelligence[edit]

Weak AI gives computers elements of human cognition--the ability to think like a human. A good example of Weak AI is computer software which uses an approach known as "brute force" to calculate probabilities. The case of chess master Garry Kasparov vs IBM's Deep Blue program is a good illustration. Although Kasparov had the advantage of human intuition, Deep Blue had the ability to calculate 200 million probabilities per second. In the end Kasparov was overcome by Deep Blue's super-efficient ability to calculate those probabilities. Another example of Weak AI is spell-checking software, which--like Deep Blue--uses probabilities to make determinations. When a writer spells a word incorrectly, the software does not check the spelling like a grade school teacher does, but compares the incorrectly spelled word against a dictionary of correctly spelled terms with similar characters. It then determines which word was probably intended by the author using a set of programmed heuristics. The efficiency of Weak AI is determined by the speed at which such comparisons can be made and the depth of the database which it draws upon.

Strong Artificial Intelligence[edit]

Strong AI claims that a computer can be given the ability to think on a level with humans, possibly becoming "aware" of itself at some point. Strong AI is still in its infancy and has yet to develop to a credible level among scholars. At Stanford University, professor Douglas Lenat and a team of programmers, linguists, theologians, mathematicians, and philosophers have recorded 1.4 million basic truths and generalities into a database in an attempt to give a computer the intelligence of a 12-year-old. They have also given the computer the ability to ask questions and form assumptions. Strong AI uses two approaches. Cyc which loads the database with facts and rules and Co.

Conclusion ::[edit]

The efforts of Douglas Lent and other teams of scholars who have attempted to impart a database with the distinctly human quality of intuition are admirable. However, the author does not see how this approach differs from Weak AI. Although the type of information differs, both methods require the building of databases and, at the core, use probabilities or high-speed comparisons to draw conclusions. Computers, though capable of performing tasks far beyond the ability of humans, are not (at this point) able to mimic human intuition, because intuition is a unique and individual possession of each person. It cannot be created or quantified by facts, figures, basic truths or generalities. It is not downloadable, regardless of the size of the data base.