New HAABI Scale Measures Human-AI Emotional Bonding in 20 Questions

A new paper on arXiv introduces the Human-AI Affective Bonding Inventory (HAABI), a 20-item, four-factor scale designed to measure emotional bonds users form with conversational AI. As AI chatbots become more affectively responsive, existing tools borrowed from human relationship psychology may miss unique aspects of human-AI bonding. The researchers conducted two studies: first, thematic analysis of semi-structured interviews with 52 emotionally engaged conversational AI users; second, a self-report inventory validated on 673 Chinese users. Exploratory and confirmatory factor analyses revealed four dimensions: emotional realism (how real the AI's emotions feel), separation anxiety (distress when not interacting), emotional investment (time and energy spent), and romantic intimacy (feelings of closeness or romance).

The scale showed good reliability, construct validity, and known-groups validity, meaning it can distinguish between users with different levels of attachment. The HAABI provides a neutral, user-centered tool for researchers studying how affective bonds form, their psychological outcomes, and potential risks. This is especially relevant as chatbots like ChatGPT, Character.AI, and Replika foster deep user relationships. The paper is publicly available on arXiv and is pending DOI registration.