A Native Intelligence Metric for Artificial Systems
John A. Horst
We define native intelligence as the specified complexity inherent in the information content of an artificial system. The artificial system is defined as a system that can be encoded in some general purpose language, expressed minimally as some finite length bit string, and decoded by a finite set of rules defined a priori. Using this definition of native intelligence, we employ a chance elimination argument in the literature to form a simple, but promising native intelligence metric. Several anticipated objections to this native intelligence metric are discussed.
Performance Metrics for Intelligent Systems, Workshop | | Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop | NIST
artificial systems, chance elimination, complexity theory, design inference, intelligent metric, linear systems, metrics, native intelligence, probability theory
A Native Intelligence Metric for Artificial Systems, Performance Metrics for Intelligent Systems, Workshop | | Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop | NIST, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=824478
(Accessed November 30, 2023)