Getting started with AI, part 2: What some planners are actually saying

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. 

Previously, I wrote about how planners can take their first practical steps with AI, regardless of technical background, agency size, or budget. From this and other conversations I’ve had with planners, it appears that their experiences vary at multiple levels. 

I wanted to explore this further, so I reached out to a handful of planners across different settings: a regional planning director, a senior program manager at a private transportation firm, a principal planner at a county transportation department, an economic development coordinator at a major city and a deputy director at a municipal innovation team. Some are AI enthusiasts who use it many times a day. One is candidly skeptical and still waiting for her own "lightbulb moment." Their answers, taken together, paint a more nuanced picture of where the profession actually is, even coming from a small sample of planners.

A few notable themes jumped out.

Everyone starts somewhere small

I sometimes get the feeling that the "AI in planning" conversation makes people feel like they are already five steps behind. The reality, even among people who now use these tools regularly, is that this is all very new and almost everyone begins with something modest.

Mateo Alexander, an economic development coordinator, now uses ChatGPT, Claude and Gemini "several times per hour" and considers AI a core part of his workflow. But his first real use was simple: data analysis. "Instead of matching thousands of data entries, I can have AI do that for me," he told me. Nneka Sobers, Deputy Director at the NYC Innovation Team, started at a similar level. Her first use was editing emails, "particularly the detailed ones that required careful attention, balancing nuance and clarity." She made what she called "a deliberate choice to start by giving it context I already knew well," and that turned out to be a good instinct. Michael Brown, a director at a regional planning organization, described his entry point in similar terms: "testing how it could support writing, framing ideas, and summarizing complex information."

In other words, none of these planners walked in with a big challenge to confront. They picked one common task and tried it. That, more than anything, is the lesson I would underline from my previous column. Pick a small, low-stakes thing on your to-do list and start there.

The "click" moment is usually about time

In nearly all cases, there was a specific moment when interest in AI shifted from curiosity to genuine utility. Those moments almost always have the same shape: a task that used to take hours or days suddenly takes minutes.

For Nneka, it was a synthesis of community engagement. Her team builds extensive engagement processes that generate stacks of sticky notes, worksheets and other materials, and turning those into structured findings is some of the most time-consuming work they do. She developed a specialized AI tool that "compresses what would have taken us weeks of synthesis to complete into a synthesis process that takes a few hours, without compromising depth." For Tara Reel, a senior program manager at a transportation firm, the click moment came while reviewing state and regional Six-Year Improvement Programs. These dense capital planning documents used to require line-by-line review. Now, he uses AI to "scan for relevant project types, funding patterns, timelines, and geographic priorities," which turns "a very time-intensive review process into a more data-driven, scalable approach."

What I find interesting is that none of these planners describe AI as doing something they could not do themselves. They describe it as doing things they can focus on for quality. That distinction matters. It reframes AI less as a replacement and more as a way to reduce the time previously spent on mechanical work.

Writing is the universal on-ramp

From all five interviews, I would say the most common entry point was writing: drafting emails, structuring stakeholder communications, refining a difficult message and getting unstuck on a first draft. Michelle Stafford, a Principal Planner in transportation who self-describes as not particularly comfortable with technological change, told me she uses AI "as a starting point in writing correspondence in a challenging situation or topic." She is clear that she never uses the verbiage verbatim, but she appreciates "a particular approach or sensitive phrasing when I get stuck."

I think that last point is important. Several people emphasized that they treat AI’s writing as a starting place, not an ending place. Michael Brown put it this way: "Use AI as a first draft generator, not a final author." Mateo Alexander described it as a "brainstorm partner." Tara Reel, the senior program manager, said he now approaches tasks "more iteratively and strategically, rather than starting from a blank page."

If you are looking for the lowest-risk way to try AI in your planning work this week, draft a somewhat challenging email with it. Not a simple one. The one you have been putting off because you do not know quite how to phrase it.

Skepticism is not the opposite of progress

Michelle Stafford’s responses were quite insightful, mostly because she described not yet having found her own breakthrough. She told me she is "generally waiting for the lightbulb moment" in her day-to-day work. She also painted a vivid picture of what AI could eventually do for her program: real-time scoring of community-nominated street improvements against more than 20 criteria, with instant feedback to the resident who submitted the application. She can see it. It’s a matter of having access to the right tool (or agent) to do it (see my previous column on AI agents). She does not have that tool yet.

I think Michelle’s perspective represents a much larger share of the profession than the AI conversation would suggest. I am personally very interested in hearing the full range of views. We are indeed at the beginning stages, and we benefit from hearing from our peers. The planners who are skeptical are often the ones thinking about situations when human judgment, relationship building and live community engagement cannot be replaced. Michelle put it well: "Soft skills may become the highest priority in this field."

Common concerns are clustering

Across very different agencies and roles, the concerns I heard were quite consistent.

Accuracy and over-reliance came up in every single response. The senior program manager called it "accuracy and over-reliance" almost verbatim. Michael Brown warned about "people using AI outputs without interrogating accuracy, bias, or context, especially in a field like planning where decisions have real community impacts." Nneka noted that AI tools "are not subject-matter experts" and that working with them requires "a kind of disciplined skepticism that doesn’t always come naturally, especially when the tool sounds confident."

Data sensitivity in government contexts came up repeatedly. So did concerns about job displacement, particularly for early-career planners. Mateo, who is among the more enthusiastic AI users I spoke with, raised concerns about the infrastructure side, including data centers and the land use questions that come with them, a topic I have covered in this column before.

None of these concerns is a reason not to start — these are reasons to start thoughtfully.

What I would tell a planner getting started today

To synthesize what these five planners told me, the advice for someone just beginning their exploration of AI boils down to a few simple points.

1. Start with the task that takes more time than you think it should. Nneka suggested, "The task that sits on your to-do list longest whose effort feels disproportionate to its outcomes. That’s usually where AI can reduce the most friction."
2. Use AI to learn AI. Mateo’s advice was direct: "Just dive in and learn. If you don’t know where to start, you can even ask AI." That recursive quality is one of the genuinely new things about this technology.
3. Keep your professional judgment at the center. Michael’s framing is one I will be borrowing: "Treat AI as a time saver, not a substitute for planning expertise." 
4. Do not overlook the value of cross-generational collaboration. Michelle’s point about pairing 20 years of experience with newer staff who bring different skill sets is one of the smartest things I've heard in this whole exercise. The most effective AI adoption in planning agencies will look like that kind of partnership, not a top-down rollout.

One more thing

I want to thank everyone who took the time to share their experience for this column. Their candor, including their hesitations, made this far more useful than any one person’s view could be.

If reading this has nudged you to take that first step, you also should know that I will soon be rolling out a class with Planetizen on getting started with AI in planning practice. Keep an eye out for it. We will go deeper into many of the themes I touched on here, with hands-on examples and time for your questions. I hope to see you there.

Special thanks to:

Michelle Stafford, Principal Planner, Arlington County Government (VA) Department of Environmental Services, Division of Transportation – Transportation Planning.

Mateo Alexander, Economic Development Coordinator on the Catalytic Development Team, the City of Dallas Department of Economic Development.

Nneka Sobers, Deputy Director on the NYC Innovation Team, a creative policy lab within New York City’s Mayor’s Office.

Michael Brown, AICP, Director, Chicago Metropolitan Agency for Planning,

Tara Reel, Senior Program Manager and Business Development Lead.