The urban right to AI: Pluralistic co-design and governance of public space
This thesis argues for a civic "Right to AI" and a pluralistic governance framework for urban public spaces, demonstrated through participatory research in Montréal that uses computer vision and generative AI to map and preserve contested community values rather than averaging them into a single objective.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Cities have always been built from two things: the physical world of streets, sidewalks, and buildings, and the invisible world of rules, maps, and decisions that tell officials how to manage them. For a long time, the invisible world was made of paper plans and human judgment. Today, that invisible layer is increasingly made of computer systems. These systems look at photos of streets, score them on how safe or welcoming they seem, and even generate new images of what a street could look like if redesigned. When these systems are treated as simple, neutral tools, the choices they make about what to measure and how to weigh different needs are hidden. They might prioritize what looks good to a standard camera while ignoring what feels safe to a person with a disability or a parent with a stroller. The result is a city where the computer's version of "good" becomes the only version that matters, potentially erasing the complex, conflicting ways different people actually experience their neighborhoods.
This is the problem a new doctoral thesis from the University of Montreal sets out to solve. The researcher, Rashid Ahmad Mushkani, argues that we must stop treating these computer systems as mere tools and start treating them as a form of civic infrastructure, just like the water pipes or power grids that run under our streets. If a water pipe breaks, the city has a process to fix it. If a computer system makes a bad decision about which street to repair, the city needs a similar process to question, correct, and update that system. The thesis proposes a "Right to AI" for public spaces, a set of rules that ensures people have a say in how these systems are built, that the systems can be challenged when they get things wrong, and that the messy disagreements between different groups of people are not smoothed over by a single computer score.
To test this idea, the researcher went into the neighborhoods of Montreal and asked a diverse group of residents to look at photos of local streets. Instead of asking them to simply rate a street as "good" or "bad," the researcher asked them to explain what made a street feel inclusive, safe, or accessible. The participants included long-time residents, newcomers, people with disabilities, and members of various cultural communities. The goal was to see if everyone agreed on what a "good" street looked like. The findings were clear: people did not agree. A street that one person saw as vibrant and lively, another might see as chaotic and unsafe. A street that felt welcoming to a young adult might feel isolating to an older person. The study found that while people could often agree on simple, visible things like whether a sidewalk was broken, they struggled to agree on complex social feelings like "inclusivity." When the researcher brought people together to discuss these differences, they could find some common ground, but significant disagreements remained. These disagreements were not mistakes to be fixed; they were real signals that different people have different needs and values.
The research then took these human insights and used them to train computer models. In one part of the study, the researcher built a system called "Street Review" that learned to predict how different groups of people would rate a street based on its photos. The system could generate a map of the city showing where different groups might feel safe or excluded. In another part, the researcher worked with community organizations to create a massive collection of images and preferences, called LIVS, to teach a computer how to generate new images of public spaces that respected these diverse values. The computer was asked to choose between different design options, and the results showed something surprising: even after being trained on thousands of human preferences, the computer often could not decide which option was better. It would remain "neutral," unable to pick a winner.
This neutrality turned out to be a crucial discovery. The researcher argues that when a computer cannot decide between two options, it is not a failure of the technology. It is a sign that the values at stake are genuinely in conflict and cannot be solved by a simple calculation. Instead of forcing the computer to pick a winner, the system should flag this uncertainty and send the decision back to the people. The thesis proposes a new way for cities to work: a cycle where computers help visualize options, but humans, organized in small, recurring groups, make the final calls. These groups would review the computer's suggestions, discuss the disagreements, and decide which streets to fix and how. They would also have the power to pause or stop a system if it starts making harmful decisions.
The work does not claim to have solved the problem of building perfect cities. It does not say that computers can replace human judgment or that we can find a single definition of a "perfect" street that satisfies everyone. Instead, it offers a practical framework for managing the tension between technology and human diversity. It suggests that the best way to use AI in our cities is not to let it make the final decisions, but to use it to make our disagreements visible and manageable. By treating the computer's logic as something that must be constantly checked, updated, and argued over, cities can ensure that their digital tools serve everyone, rather than just the majority or the most visible groups. The research concludes that a city that wants to be truly inclusive must build a system where the right to question the computer is as important as the right to walk on the sidewalk.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.