This paper makes a distinction between measurement at surface and deeper levels. At the deep levels, the items measured are theoretical constructs or their attributes in scientific theories. The contention of the paper is that measurement at deeper levels gives predictions of behavior at the surface level of artifacts, rather than just comparison between the performance of artifacts, and that this predictive power is needed to develop artificial intelligence. Many theoretical constructs will overlap those in cognitive science and others will overlap ones used in different areas of computer science. Examples of other sciences of the artificial are given, along with several examples of where measurable constructs for intelligent systems are needed and proposals for some constructs.
Proceedings Title: Performance Metrics for Intelligent Systems (PerMIS 01)
Conference Dates: September 3-7, 2001
Conference Location: Mexico City, -1
Pub Type: Conferences
artificial intelligence, deep measurement, intelligence, intelligent systems, machine intelligence, measurable constructs, measurement of performance, surface measurement, theoretical constructs