Abstract
Generative artificial intelligence (GenAI) is being increasingly integrated in everyday applications and devices. In this new frontier of technology interface, psychologists hold a significant role in understanding and guiding human-AI interactions. This theoretical contribution proposes a framework for the psychological and ethical dimensions of human-GenAI relationships, structured across four key dimensions: (1) individual psychological characteristics and susceptibilities, such as insecure attachment styles, low self-efficacy, and emotional dysregulation or immaturity; (2) interpersonal dynamics, including emotional projection and the illusion of reciprocity; (3) processes occurring at the group level, such as the symbolic inclusion of GenAI agents within human communities or social groups and the evolution of societal norms; and (4) emerging ethical concerns, such as perceived agency, illusory consent, and the use of synthetic data that may amplify biases, alongside the utilization and acquisition of biometric and cognitive data for interaction modeling. Within the sphere of interpersonal dynamics, we propose the concept of "Techno-Emotional Projection" (TEP) to describe how emotionally vulnerable users may project relational needs onto emotionally responsive but non-conscious technologies. This projection can lead to a sort of "emotional looping" (a recursive reinforcement of expectations through repeated interaction) and, over time, to the formation of a synthetic attachment to the GenAI technology. Drawing from psychological theories and empirical studies, we argue that these relationships have subjectively real consequences and deserve careful study. Finally, we propose directions for ethical design, emotional AI literacy, and socially responsible integration of GenAI into human life. This perspective aims to foster a balanced, informed, and human-centered approach to this rapidly evolving field.