Intelligence: information use and adaptive capability

Abstract

Intelligence depends on how a system makes information usable, retains it and changes through experience. This perspective connects five established dependencies across biological and artificial systems: the conditional value of information, access through available operations, persistence relative to future demands, coupled changes through shared organisation, and the effects of present operations on later capability. A running example of a learner using an AI assistant and saved solutions shows how these relationships jointly structure a practical question: when does assistance improve both present performance and what the learner can do later? Restricted mathematical comparisons distinguish learning opportunities from learning signals and encoding allocation from later recall. An extension based on published memory models explains how expected assistance can redistribute encoding effort and why reduced total investment requires an avoided cost or alternative use. Biological studies, human learning research and a bounded neural forecast illustrate the relationships at different levels of evidence. The account treats intelligence through graded capacities and allows implementations to differ. Its contribution is an integrative perspective with reproducible comparisons. Supporting materials include mathematical derivations, qualified evidence records and reproducible analytical code. Licenses. Paper, supplement, figures and original evidence materials: CC BY 4.0. Accompanying code: MIT. Third-party works retain their own terms.