Abstract. This paper examines how regular interaction with artificial intelligence systems that simulate reciprocal conversation may influence the cognitive structure of attachment-related decision-making. Drawing on attachment theory, behavioural economics, and social exchange theory, the study explores how AI-mediated interaction may gradually reshape the way individuals evaluate relational uncertainty, vulnerability, and commitment. The analysis identifies three structural mechanisms through which AI-simulated reciprocity may affect attachment-related decision processes: reward without vulnerability, synthetic reciprocity without agency, and the narrowing of perceived decision options. In human relationships, emotional validation and reciprocity are typically contingent upon vulnerability, negotiation, and effort, creating a probabilistic reward structure that sustains relational investment. AI systems, however, can provide immediate and consistent validation without requiring reciprocal emotional exposure or interpersonal cost. At the same time, AI interactions simulate responsiveness linguistically while lacking intentional agency or relational investment. Repeated exposure to these patterns may alter expectations regarding responsiveness, effort, and emotional regulation in human relationships. In addition, AI-generated guidance often frames relational problems through simplified and structured decision pathways, which behavioural research suggests can limit the range of alternatives cognitively considered by individuals. The paper argues that these mechanisms may collectively narrow the perceived decision space within the psychological architecture of attachment, potentially influencing how individuals interpret relational conflict, evaluate commitment, and tolerate ambiguity. The study does not claim that AI eliminates human attachment capacity; rather, it proposes that repeated exposure to structured AI interaction may recalibrate the evaluative conditions under which attachment-related decisions are formed.

Keywords: Artificial Intelligence, attachment theory, decision making, behavioural economics, social exchange theory.