Artificial intelligence and the psychology of human connection

The paper introduces the machine-integrated relational adaptation (MIRA) model, a middle-range theoretical framework describing when, how and why AI comes to function as a relational entity rather than a tool. MIRA separates two roles: relational partner, where the user treats the system as a quasi-social agent in direct interaction, and relational mediator, where the system sits between two humans and reformulates what passes between them. The framework is organised as antecedents, mechanisms, moderators and outcomes, and draws its explanatory apparatus from attachment theory, social exchange theory and epistemic trust rather than from AI-specific scholarship.

Four mechanisms carry the relational effect: linguistic reciprocity, psychological proximity, interpersonal trust, and relational substitution versus enhancement. The authors ground all four in verbal behaviour on the argument that language is currently the primary interface through which people encounter AI, that words function as both signal and mechanism of social cognition, and that a century of empirical work on language and social process makes falsifiable prediction possible now. The framework treats trust as part of the interaction loop rather than only an outcome, on the basis that a system which has accumulated relational legitimacy through thousands of affirming exchanges also carries disproportionate persuasive weight.