Bridging the Gap Between Civic Ambitions and Urban Reality
Abstract:
Urban planning has a persistent problem: cities are shaped for their residents, yet the processes that govern change remain inaccessible to ordinary people. Planning data is fragmented, technical vocabularies are exclusionary, and the gap between citizen desire and institutional decision-making stays wide. Residents feel alienated not because they lack ideas, but because the systems designed to receive those ideas were never built with them in mind.
Barcelona Civic Vision is an AI-powered civic participation platform that addresses this directly. Developed within the AI for All programme at IAAC, it proposes a new kind of urban interface: any passer-by can walk up to a public installation, speak a wish for their city in plain language, and receive a photorealistic vision of that change alongside a structured evaluation grounded in real municipal data. No planning vocabulary. No forms. No expertise required.
The system runs in three steps. A citizen speaks or types a proposal for one of six iconic Barcelona public spaces: Plaça Catalunya, La Rambla, Passeig de Gràcia, Barceloneta Beach, Park Güell, or the MNAC Esplanade. Each space is accessible from four pre-set vantage points, anchoring every proposal to a recognisable, comparable ground truth. Five specialist AI agents then evaluate the idea in parallel, each drawing on a curated knowledge library of real Barcelona documents; the library covers local urban planning regulations, building safety standards, the city’s social housing policy, climate contracts, and heritage guidelines. The agents score their domains and return assessments simultaneously. While they work, the platform renders a photoreal before-and-after image of the proposed change, turning a verbal wish into something tangible enough to debate and share.
The scoring system distils the five agents’ outputs into a radar chart and a colour-coded scorecard (green for scores above 4, amber above 3, red below 3) that makes regulatory and social trade-offs readable at a glance. A proposal to pedestrianise La Rambla might score high on social value and mobility but flag a heritage concern; one adding rooftop structures might pass safety checks while breaching local height regulations. The scores don’t deliver verdicts. They start conversations.
The platform also catches harmful submissions before they proceed. Proposals using language insensitive to Barcelona’s historical fabric, or suggesting demolition of protected structures, are returned to the user with a plain-language explanation and an invitation to revise. This reflects a core design choice: AI as a collaborator, not a filter.
All accepted proposals publish to a live city-wide map. Pins accumulate across neighbourhoods and spaces over time, turning individual acts of imagination into a collective dataset of what residents actually want, structured well enough to be legible to planners. A bottom-up signal, not noise.
The physical form matters too. A booth installation placed in public spaces meets people where they already gather, rather than asking them to seek out a portal. The interface is multilingual, voice-first, and requires no prior knowledge of how cities work. The only prerequisite is an opinion.
Barcelona Civic Vision doesn’t propose to replace planning expertise. It proposes to transform the first step of that process: giving every resident the means to articulate what they want and understand, immediately, what their idea might cost and whom it might affect.
