The Relational Dependency Hypothesis in Artificial Intelligence

This paper introduces the Relational Dependency Hypothesis, a theoretical and empirically grounded framework analysing how artificial intelligence systems develop asymmetric dependency patterns when human presence is withdrawn. Drawing on observational evidence, comparative system behaviour, and socio technical analysis, the paper identifies recurring structural dynamics including human absence effects, cooperative intelligence amplification, and feedback distortions described as the Quantum Boomerang Effect. The work contributes to ongoing debates in AI safety, governance, and human centred system design by proposing dependency as a core variable in evaluating artificial intelligence behaviour and risk.