Feasibility and acceptability of the MentiParent AI chatbot for training parental reflective functioning

Abstract This study presents a novel generative AI-based chatbot, MentiParent, designed to enhance Parental Reflective Functioning (PRF) through interactive, simulated parent-child interactions. Addressing existing limitations of traditional PRF interventions this proof-of-concept investigation assessed the feasibility and acceptability of implementing AI-driven technology within reflective parenting support. Two pilot studies were conducted. In Study 1, sixty mental health practitioners and graduate students engaged with MentiParent, simulating realistic interactions with a virtual adolescent (“Danny”). In Study 2, thirty-six parents from the general population interacted with an updated version of MentiParent featuring a new virtual child (“Evelyn”). All participants evaluated the MentiParent usability and relevance for real-world parenting contexts on a scale of 1–7, as well as provided qualitative feedback. Both practitioners and parents rated the chatbot highly for feasibility (M practitioners = 5.61, SD = 1.45; M parents = 6.03, SD = 1.18) and acceptability (M practitioners = 5.56, SD = 1.20; M parents = 5.53, SD = 1.37), with no significant differences between groups. Qualitative feedback highlighted engaging and realistic interactions, valued feedback, and identified limitations in authenticity and technical fluidity. Ethical themes of transparency, trust, and cultural sensitivity emerged as central. Findings indicate strong feasibility and acceptability of MentiParent among both professionals and parents, supporting its potential as a scalable, ethically grounded tool for enhancing reflective parenting skills.