In Planning, the Possibilities With AI are High. So are the Stakes.

This article is part of an ongoing column on AI and planning by urban planner and AI expert, Tom Sanchez. Learn more about Tom and read more installments of his column.

Artificial Intelligence (AI) is moving quickly within the urban planning profession, and with it comes both anticipation and unease. For planners, the stakes are tied to the communities we live in, the neighborhoods we work with, and the trust people place in our decisions. The choices we make about how to use AI will shape not only the kinds of insights we gain from data, but also the public's trust in planning. Planners risk undermining that trust and repeating past mistakes. That’s why this conversation matters now, before AI becomes just another everyday tool in planning practice.

Back in 2013, a group of MIT researchers showed just how fragile “anonymous” data really is. They analyzed a dataset of 1.5 million people’s cellphone location records over 15 months, with all personal identifiers removed. By cross-referencing the anonymized records with just four points of outside information — such as the approximate times and places someone was observed — they could re-identify 95 percent of individuals in the dataset. It was a wake-up call: data that feels safe and anonymous often isn’t. And with AI today relying on huge amounts of this kind of information, the lesson feels even more urgent.

For planners, this isn’t some distant academic issue. Cities and planning departments are already turning to AI to sift through streams of data — whether it’s from smartphones, transit passes or connected vehicles — to better understand how people move through neighborhoods. That can be powerful, but as the MIT case reminds us, it can also slide into dangerous territory if we’re not careful. Privacy, bias, and accountability aren’t optional extras, they’re the guardrails that keep AI from doing harm while we try to do good.

Credit: Tom Sanchez

The ethical stakes in planning AI

AI holds a lot of promise for planning. It can make information more accessible to more people, help us spot patterns we might miss, and give us better tools for making decisions. But there’s a flip side. Without the right safeguards, these same tools can deepen inequities, invade people’s privacy, and chip away at public trust. Planning has always been about working in the public’s best interest. When things go wrong, (redlining is a well-known example) the impacts can linger across entire communities for generations.

And now we’re facing new challenges with generative AI. These tools can spin up realistic images, stories, or even data, which sounds exciting but also opens the door to disinformation and manipulation. The data behind these systems can carry hidden biases that shape how neighborhoods and communities are portrayed. Automated content might also put speed over authenticity, drowning out real voices in the process. On top of that, many of these systems are so opaque that residents can’t tell how a result was created, let alone push back if it feels wrong. If planners aren’t careful, these technologies could end up reinforcing old power imbalances and making the planning process less transparent, less participatory, and less fair.

Privacy: Lessons from smart city data collection

One of the clearest lessons about privacy in planning comes from Toronto’s Sidewalk Labs project. The idea back in 2017 was to build a “smart neighborhood” with sensors tracking everything from how people walked down the street to how much energy they used, even down to their trash. On paper, it sounded innovative and efficient. But the more people learned about it, the more uneasy they became. Residents started asking tough questions: Who controls all this data? Who makes money from it? And do people really get a say if they’re being tracked every time they step outside? In the end, the project fell apart within just a few years, largely because privacy and accountability concerns were never resolved in the public’s eyes.

A rendering of Toronto’s Quayside smart city. Credit: Sidewalk Labs

For planners, the message is pretty clear, privacy can’t be an afterthought. It has to be built into projects from the very beginning. That might mean setting up independent groups to safeguard data, adopting local rules that spell out people’s rights, or making sure community conversations about new projects include honest talk about surveillance and consent. And sometimes it’s about asking a simple question: do we really need all this data in the first place? The Sidewalk Toronto story reminds us that public trust is as valuable as any dataset, and once it’s gone, it’s incredibly hard to win back.

Bias: Identifying and mitigating systemic inequities

Bias in AI isn’t just a glitch in the code, it often reflects the inequities we’ve lived with for decades. If past infrastructure spending poured more resources into certain neighborhoods, an AI model trained on that history might suggest doing the same again. In other words, the technology can quietly carry forward old patterns of unfairness. And bias can sneak in at many points: through the data we choose, the algorithms we use, or even the way we interpret the results.

The good news is there are ways to push back against this. Planners can start by taking a hard look at the data before it ever gets fed into a model, asking whether it’s complete and whether it truly represents all communities. They can also bring residents into the process, giving people a chance to question both the inputs and the outputs. And just as important, planners need to be transparent about how decisions are made, including the limits of the AI tools they’re using. Being open about what a system can and can’t do helps guard against the tendency to treat its recommendations as unquestionable truth.

Credit: Tom Sanchez

Accountability: Defining responsibility in AI-driven decisions

Accountability gets tricky when planners use AI tools built by outside companies. If something goes wrong, who’s really responsible, the vendor who built the system, or the planning department that relied on it? Without clear rules in place, it’s easy for things to fall through the cracks. And on top of that, there’s a natural tendency to trust whatever the computer spits out, even when we shouldn’t.

That’s why keeping humans firmly in the loop is so important. For planners, that means a few practical steps: keeping records of how AI tools were used in decisions, making sure contracts with vendors require transparency, setting up internal review groups to look at AI projects, and making sure staff are trained to ask tough questions about the results. At the end of the day, AI should support planning, rather than replace the responsibility planners have to the public.

Integrating ethics into planning practice

If planners are going to use AI, ethics can’t be something we tack on at the end, it has to be part of the process from day one. That means asking hard questions right up front: does the data leave out certain neighborhoods or groups? Are there risks hidden in how it’s collected or used? Once a project is underway, it’s not enough to just let the system run, we need to keep an eye on the outputs, involve the people affected, and make sure community voices are part of interpreting what the AI is telling us. And the work doesn’t stop once the tool has been put to use. Planners should treat evaluation as an ongoing, public process. 

Independent reviews and open reporting can help spot problems early and keep things on track, even if that means slowing down or hitting pause. Some cities are already taking this seriously by forming ethics committees that bring planners, technologists, lawyers, and community members together to keep watch. The big idea is simple: technology in planning should always protect people’s rights, promote fairness, and serve the public good.

A call to action

As planners, we play an important role in deciding whether AI becomes a tool that strengthens democracy or one that quietly erodes it. The examples we’ve seen, from re-identified “anonymous” data to failed smart city projects, remind us that trust is both fragile and essential. If we build privacy, fairness, and accountability into AI from the start, these tools can expand participation, make our work more transparent, and help communities feel heard. If we don’t, we risk repeating old mistakes with new technologies. The choice is ours, and the time to set the right course is now.

Resources:

Course: Navigating the Ethical Landscape for AI in Urban Planning

Paper: The Ethical Concerns of Artificial Intelligence in AI

—

I invite every planner to follow this column each month and to offer your feedback, ask questions, and share your own experiences. I would love to hear about how you are currently using AI, the challenges you are facing, and the ideas you would like to explore. If you have a question that you would like answered, please email [email protected] with “AI question for Tom Sanchez” in the subject line. Let’s learn together how to put AI to work in ways that truly serve our communities.