Getting Started with AI: A Practical Guide for Urban Planners

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

Over the past nine months in this column, I’ve touched on a range of topics: why planners can’t ignore AI, how artificial neural networks work, why representative data matters more than ever in an AI-driven world, how AI can help (or hurt) community engagement, what digital twins look like, whether planners might lose their jobs to automation, and what the data center boom means for our communities. 

Those are big topics. But when I talk to planners, one question frequently comes up about AI that is not related to any of these: “But how do I actually get started?”

It’s a good question, and it often applies to any new technology. So let’s step back from the big picture and talk about the practical first steps that any planner can take, regardless of technical background, agency size, or budget.

1. You’re probably already using AI

As I mentioned in my first column, most planners have been using some form of AI for a long time now. Spell check, grammar tools, and a variety of statistical methods are all forms of AI, just wrapped in interfaces we’ve grown comfortable with. More recently, planners are already using tools like ChatGPT, Gemini, or Claude to draft emails, summarize reports, or create graphics. If you’ve done any of these things, you’ve already started.

The point I want to make here is that getting started with AI is not about making a dramatic leap. It’s about being more experimental and intentional with the tools already within reach and gradually building your confidence and understanding from there.

2. Start with what you do every day

The best way to begin is to look at your own daily workflow and identify the tasks that are repetitive, time-consuming, or involve processing large amounts of text or data. These are the places where AI could provide immediate value. Think about the staff report that takes two days to draft, the stack of public comments that requires a lot of time to read closely, or the demographic data in a spreadsheet, which needs to be made into a usable summary.

Here are some concrete starting points:

First, try using a large language model (LLM) such as ChatGPT, Claude, Copilot, or Gemini to summarize a document. For example, first ask the AI to summarize the main themes. But then give it a role-playing prompt: “You are a city planner for the town of X working on Y (e.g., a zoning code revision) and need to summarize these public comments into their main themes.” And then try a more specific prompt that is more informational or is an announcement for a public meeting on a particular project. Most people will also be surprised to observe how the AI takes on the role you’ve described and helps to add context to the prompt.

Second, use AI to help you draft, not write, routine communications. Spend time crafting emails to co-workers, community members, or elected officials; try giving an AI tool some bullet points and asking it to draft a professional message. Then edit it. You’ll save time and, over a few iterations, develop a sense for how to “prompt” effectively. This is a very important skill and an important part of the process.

Third, experiment with data exploration. Many planners work with Census data, permit records, or traffic counts, but don’t have time to dig too deeply into the numbers. Tools like ChatGPT’s data analysis features (or, if you are willing to learn a bit of Python, free tools like Google Colab) can help you quickly visualize trends, run correlations, or spot outliers that could take a significant amount of time to generate in the usual ways.

3. You don’t need to be technical, but being curious is important

One thing I have mentioned when I talk about AI capabilities is that you do not need to know how to code to use AI effectively. What you need is curiosity and a willingness to experiment. The learning curve for tools like ChatGPT is remarkably flat: first, you type a question (aka a prompt) and get an answer. The real skill lies in learning to ask better questions, evaluate answers critically, and apply them in a planning context. This is often referred to as the “workflow.”

That said, I would encourage any planner who is curious about AI to invest some time in understanding the basics of how these tools work. You don’t need to build a neural network, but knowing what an LLM does, why it sometimes makes things up, and what kinds of tasks it’s well-suited for versus those it’s not can make you a more effective user. That’s why I wrote my earlier columns on neural networks and on the stakes of AI in planning, not to turn planners into engineers, but to give you enough vocabulary to ask some questions and participate in discussions about how these tools can be used in your department.

4. Build a small pilot, not a grand strategy

If you’re a planning director or manager thinking about bringing AI into your organization, my advice is to start small. 

On this point, I asked COMPASS MPO Executive Director Craig Raborn for his suggestions. At COMPASS, he started with a staff survey asking whether they were already using AI, whether they were interested in using it more, what concerns they had about using it, etc. Through the survey and subsequent one-on-one discussions, he identified the early adopters and those likely to be late adopters. He had some of the early adopters test uses and sought to get the later adopters more comfortable with AI. 

Also, don’t try to begin with a comprehensive AI strategy. Craig recommends picking one well-defined problem, assigning a small team (or even just one motivated staff member), and giving them a day or two to test an AI approach to that problem (e.g., analyzing recent crash trends). He also suggested tasking someone new to planning or a specific topic with using an AI approach and giving them time to consider whether the AI is providing them with good information.

The goal isn’t to deploy a finished product. It’s to experiment and learn. What does the tool do well? Where does it fail? What surprised you? What made staff uncomfortable? The answers to these questions may teach your organization more about AI readiness than any training video.

5. Know the risks

As I’ve written in previous columns, AI can do many things, but it is not magical. Language models can and do produce inaccurate information, often with confidence. They can reflect biases embedded in their training data, which is especially dangerous when we’re making decisions that affect vulnerable communities. And they raise real questions about transparency: if a planning recommendation was shaped by an AI tool, do residents and elected officials have a right to know?

My suggestion is to start with a simple disclosure practice run. Pay attention to how stakeholders respond and note the kinds of questions they ask. Strict policies in the early stages may constrain testing and learning, when the preference may be to experiment. If AI helped you draft, summarize, or analyze something, say so. This can build trust rather than eroding it, and it positions your agency as thoughtful and transparent about how it uses new tools. Related to this, Craig Raborn also mentioned a potential challenge: whether to focus on a platform or two and be selective in the process. As I noted in my column on community engagement, the goal should always be to use AI to hear from more people, not just to go faster.

6. Take advantage of learning resources

The good news is that you don’t have to figure this out alone. The American Planning Association (APA) has been developing resources and guidance on AI. Also, Planetizen offers several courses (including several taught by me) that help planners get up to speed, from demystifying AI terminology to hands-on workflows with generative AI tools for public participation, ethics, and more. There are many LinkedIn Learning and YouTube videos on a wide range of related topics. My book, Artificial Intelligence for Urban Planning (Routledge, 2025), walks through some basic concepts and practical applications in a way that’s meant for planners, not computer scientists. And this column will continue to be a place to explore these topics together each month.

Beyond formal resources, I’d encourage you to talk to your planning colleagues. Find out who in your office or network is already experimenting with AI. Some of the best learning happens informally, over coffee, when someone shows you how they used Claude to cut a three-day task down to an afternoon. That kind of peer-to-peer knowledge sharing is how adoption actually spreads in a profession like ours.

The time to start is now

If there’s one thing I would like you to take away from this column, it’s that getting started with AI is not as hard as it might seem, and it will most likely become increasingly important over time. The tools are accessible, the learning resources are growing, and the profession needs people who understand both planning and AI well enough to keep the technology on track. If you haven't already, I suggest that you start exploring, asking questions, and bringing your planning expertise to the table.

As always, I invite you to share your own experiences, questions, and feedback. Send me a note or leave a comment in the comment section below. I’d like to hear how you got started. 

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.

–

I want to thank Craig Raborn for his great input and feedback on this column. Craig is the Executive Director of the Community Planning Association of Southwest Idaho (COMPASS).