GenAI for Sustainable Mobility Behaviour Change: MUV's AI-Driven Game Engine

This paper presents a Generative AI framework developed by MUV (University spinoff) to address the attitude-behaviour gap in sustainable urban mobility. Anchored in Self-Determination Theory (SDT), the system uses behavioural, contextual, and qualitative data to generate personalised mobility challenges fostering intrinsic motivation—autonomy, competence, relatedness—beyond generic extrinsic rewards. The framework features a three-layer data architecture integrating user profiles, evidenced behaviours, and psychological assessments (Big Five, Hexad player typology) with open-source LLMs. A virtual mobility trainer provides real-time feedback, adapting goals to user progress and urban context. Ethical safeguards include federated learning for privacy, monthly bias audits, and transparency dashboards. The framework is being pre-tested at the Italian Institute for Environmental Protection and Research with 600 users, using A/B testing and mixed-methods evaluation to measure engagement, CO2 reduction, and inclusivity. Results will inform scalability across cities under the SDG11. Di Dio et al., 2022 (DOI:10.54941/ahfe1002741); Schillaci et al., 2024 (DOI:10.1007/978-3-031-82714-3_5).