RESPONSIBLE AI IN URBAN DESIGN: A FRAMEWORK FOR INCLUSIVE AND INTEGRATED PLANNING

Urban artificial intelligence increasingly influences which neighbourhoods receive investment, whose needs enter design briefs and how competing spatial claims are evaluated. Yet technical performance offers an incomplete basis for public decisions. This article develops a governance framework that connects AI-assisted urban design with distributive assessment, meaningful participation and accountable implementation. A targeted narrative review of eleven research and institutional sources published within the period from 27 October 2024 to 27 October 2026 is combined with a transparent numerical illustration. The synthesis distinguishes evidence of technical capability from evidence of public benefit and identifies four recurring problems: uneven representation, uncertain transferability, fragmented responsibility and evaluation through narrow performance indicators. The proposed framework links problem definition, data review, scenario generation, public deliberation, authorisation and monitoring through a documented decision record. An illustrative allocation exercise shows why aggregate accessibility gains can conceal small benefits for disadvantaged residents: imposing a minimum benefit requirement changes the preferred alternative even when the aggregate score falls. The article specifies indicators, institutional responsibilities and an evaluation design for testing this proposition in municipal practice. Its contribution is a procedure for making normative choices visible and contestable, it does not establish that the framework has already improved outcomes. Responsible adoption requires evidence that affected groups benefit, that officials can explain and revise decisions, and that the cost of oversight is included in implementation budgets.