Wednesday, July 11, 2012

Omniscience vs Rationality

Just came across an interesting concept in artificial intelligence.
Usually when we think of omniscience, we're thinking of an attribute of God a la Anselm.  But in artificial intelligence, omniscience has a different definition.

An omniscient agent knows the actual effects of its actions.


In other words, any AI program that knows the exactly what will happen for all its possible actions is omniscient.  By this definition, a good Tic-Tac-Toe AI is omniscient because it always knows the outcome.

By contrast, a rational system tries to get the best possible outcome given limited knowledge.  A chess AI would be a good example of this.  It's simply not possible with current tech to foresee every possible outcome or know the exact result of any given move.  But an AI can use its knowledge base to make rational decisions and still beat Garry Kasparov.

Further Reading:
http://informatik.unibas.ch/lehre/fs09/cs205/folien/02Agents.pdf

2 comments: