Public milestones in artificial intelligence often rely on absorbing vast amounts of collective human effort. A machine's ability to process complex queries may depend heavily on the structured information humans have already organized for themselves. When a system appears to demonstrate independent knowledge, it is frequently reflecting the massive scale of human curation beneath it.

Wikipedia recorded for February 16, 2011, that IBM's Watson artificial intelligence program won on the U.S. quiz show Jeopardy!. The program defeated Brad Rutter and Ken Jennings, the series' most successful contestants. Watson used Wikipedia, among other sources, as its knowledge base.

Demonstrating Capability

A public demonstration of this kind may serve aims beyond research. The use of Wikipedia, an openly edited encyclopedia, might point to a broader pattern in automation: such systems may depend on large collections of material that people have already organised.

The format of a quiz show provides a rigid framework where answers are definitively right or wrong, which may make it a useful testing ground for natural language processing. Success in this arena might indicate that the primary challenge for systems of this kind is not reasoning, but parsing the nuances of human phrasing to locate the correct data points.

Machines may appear most intelligent when they echo human ways of organising information. Whether a system like this understands what it processes is a debated question; one view is that it maps relationships between pieces of text very efficiently. If so, fast pattern matching could be mistaken for comprehension, and the human effort behind the underlying material might be easy to overlook.

Artificial intelligence may be less about generating independent thought and more about the rapid, precise retrieval of human consensus.

Sources