Designing for Long-Term Impact: A Hybrid Intelligence Case Study in AI-Assisted Mentalizing-Based Therapy

Le résumé fourni par la source

AI deployment in customer-facing service sectors increasingly bypasses domain experts in favor of direct automation, extracting tacit professional knowledge without returning value to the profession. We propose a two-layer Hybrid Intelligence design framework built around two mutually constitutive questions: how will building this tool be good for the domain expert, the profession, and the world; and how do we build an interaction in which the domain expert learns more from the system than the system learns from the expert. The long-term vision grounds design by building psychological safety and co-creation willingness; concrete interface elements in turn inspire and concretize that vision. We illustrate the framework through MentaBot, an AI chatbot co-developed with a clinical psychologist for Mentalization-Based Therapy, showing that a process grounded in long-term professional and planetary vision, structured expert co-creation, and metacognitive interface design produces an interface that meaningfully outperforms generic large language models — not because the underlying model is more powerful, but because the human expertise shaping it was more fully brought to bear.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.