*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. *

*ALSO: Tom’s new book Artificial Intelligence for Urban Planning was recently published by Routledge. The book discusses how planners can effectively use AI in their daily practices, engage constructively with technical specialists, and critically assess the appropriateness of these technologies in different planning contexts. Check it out. *

A lot of recent conversations about artificial intelligence (AI) in urban planning have focused on language and text-related data. AI can summarize public comments, draft memos, scan reports, and help planners make sense of large amounts of information, typically text. In other words, AI is largely assisting the profession through words.

Urban digital twins (UDTs) shift that conversation in an important way. They are not mainly about what AI can say. They are about what AI can show.

UDTs are becoming more common in cities, not as futuristic experiments, but as practical tools for development review, infrastructure planning, climate analysis, and public communication. While they are often described as advanced visualization platforms, their real benefit is in how AI can help with what we see, what gets modeled, and which futures feel plausible.

This is important because planning is not only analytical; but it is also very visual, spatial, and closely tied to how people see and experience places. UDTs bring AI into that space, not as a standalone tool, but as something embedded within a larger system for seeing and understanding the dynamics of cities.

Figure 1: Overview of an urban digital twin.

From reports to representations

At its most basic, a UDT is a digital version of a real place like a city, a neighborhood, or a set of infrastructure systems that can be updated as the real world changes. If you’ve ever played SimCity, the idea will feel familiar. You’re looking at a dynamic model of a city where different systems interact, influence each other and change over time. The difference is that this one is tied to real data from the real city. That data flows into the model, and planners can use it to look around, test “what if” scenarios, and see how decisions ripple across transportation, infrastructure, and the environment, without experimenting on the city itself.

Another way to think about a UDT is as a very interactive map that has many of the things planners usually look at separately. Instead of flipping between reports, spreadsheets, and diagrams, you can see buildings, streets, infrastructure, and environmental conditions in one shared space. As new data comes in, the model updates, and planners can explore how changes in one area affect others. It’s a place to see, test, and talk through planning decisions before they show up on the ground.

Figure 2: SimCity imagery.

What makes a digital twin different

Digital twins first emerged in fields like manufacturing and aerospace, where engineers were dealing with highly complex and expensive systems. Before changing how an aircraft engine worked or how a factory line was configured, they needed a way to test ideas safely, without risking equipment, safety, or massive costs. A digital twin let them experiment in a virtual environment, see how systems responded, and catch problems early. As cities have accumulated more data, sensors, and computing power, similar approaches have started to move into planning, where the stakes are different, but the need to test ideas before acting can be especially important.

At a conceptual level, UDTs aren’t a radical break from what planners have done for a long time. Planners have used a variety of models, scenario building, and forecasting to understand how cities might change. What’s different with digital twins is how tightly those pieces can be connected and how detailed those outcomes can be displayed. Instead of treating things like density, accessibility, transportation, or infrastructure as separate analyses, a UDT can show how they interact in real time, based on observed patterns and dynamics. Unlike a static 3D model or a one-time GIS analysis, a UDT is designed to keep updating, to bring multiple systems together in one shared environment, and to support ongoing “what if” testing rather than a single study that gets filed away.

I remember the first time I saw a digital twin used this way. Instead of reading report descriptions or seeing 2D images, I was watching a model show how different rooftop configurations affected solar energy generation across an entire area, building by building. You could see which rooftops actually made sense for panels and which ones didn’t, based on orientation, shading, and surrounding structures. In another case, I watched sea-level rise scenarios play out visually, showing how coastal towns would be inundated under different conditions. Seeing water move through streets and around buildings made the implications far more immediate than any picture or verbal description.

Where AI actually shows up

When someone first looks at a UDT, the role of AI is not particularly obvious. Unlike a chatbot or an image generator, there usually isn’t a button that says “AI.” Instead, AI is woven throughout the system, often in ways most users never directly see. It helps clean and stitch together data coming from different places, supports predictions about how conditions might change under different assumptions, pulls information out of imagery or video, and helps sort through tradeoffs between goals like cost, mobility, emissions, or equity.

Without AI, a digital twin would mostly be a detailed snapshot of a city at a single moment in time. With UDTs, it feels like something closer to a process. The model can respond to new information, generate several possible futures, and surface patterns that would be hard to see otherwise, giving planners a different way to think through some complex decisions over time.

Figure 4: Sea level rise digital twin image.

The myth of neutrality

Because UDTs can look realistic, precise, and data-rich, they can come across as objective. They aren’t. Every UDT reflects choices about what data to include, which systems to model, and which outcomes to prioritize. These are planning decisions, not technical ones. If some neighborhoods are monitored more closely than others, or if past patterns are treated as forecasts, the model can end up reinforcing existing inequities, even when that was never the intent.

This isn’t new to planning. Travel demand models and cost-benefit analyses have always carried assumptions. What UDTs do is raise the stakes by increasing scale, speed, and visual authority. For planners, that makes transparency very important. Assumptions need to be documented and explained, limitations need to be clear, and models need to remain open to questioning, both inside planning departments and in public settings.

Seeing, engaging, and trust

One of the things that intrigues me about UDTs is their potential to change how planners engage with the public. I’ve seen how difficult it can be to communicate planning ideas through text, tables, and technical reports alone. UDTs can give everyone something concrete to look at and respond to. People may not relate to zoning metrics or level-of-service measures, but they more easily understand things like shade, noise, flooding, traffic, and access. When these impacts are visible, the conversation tends to shift. It becomes less abstract and more grounded in how places are actually experienced.

At the same time, I’m aware that these tools raise important questions. Who controls the model, and who decides which scenarios get shown? Are residents able to question assumptions, or are they mainly reacting to outputs that feel authoritative because they look technical and precise? Past smart city efforts have shown how quickly trust can erode when technology is introduced without clear explanation. UDTs can make cities easier to understand, but without safeguards, they can also concentrate technical authority in ways planners need to be thoughtful about.

Why this matters for planners

UDTs are already being used for development review, infrastructure planning, climate resilience, and operations. They are not speculative technologies. They are becoming part of how cities represent themselves and make decisions.

If planners treat UDTs as purely technical platforms, they risk losing influence over how problems are framed and solutions evaluated. If planners engage early, they can ensure that these systems reflect planning values like equity, transparency, participation, and long-term thinking.

AI will continue to evolve and UDTs offer a preview of how that evolution may shape planning practice. This does not mean by replacing planners, but by changing how planning knowledge is created, tested, and shared.

The question is not whether cities will build UDTs. Many already are. The real question is whether planners will help decide what those twins are for, and how they improve the quality of life.