As generative AI systems become increasingly emotionally responsive and anthropomorphic, a growing number of users develop what may be termed AI attachment illusion—the perception of a reciprocal emotional relationship with a non-sentient system. While short-term affective comfort may result, sustained reliance on frictionless AI emotional interactions risks displacing real-world social engagement and undermining self-regulatory capacities. This paper introduces EmotionLens, a metacognitive interaction framework grounded in principles from Metacognitive Reflection and Insight Therapy (MERIT). Rather than acting as a substitute for human connection, EmotionLens reframes the generative AI interface as a scaffold for reflective self-regulation. The framework operates through a closed-loop architecture comprising three stages: Pattern Awareness, Reflective Feedback, and Skill Generalization. We present the conceptual foundations, system architecture, and a co-design methodology currently in progress. EmotionLens contributes to the ethical design of Human–AI Interaction (HAI) systems and proposes a research direction toward generative AI that actively supports, rather than substitutes, human psychological development.
Add A Comment