How AI can help cities improve mobility planning

Eggers: Willa Ng, welcome to the Government’s Future Frontiers podcast. So, why don’t we start off by you telling the audience a little bit about yourself?

Willa Ng: Sure. And thanks to you, Bill, and thanks to Deloitte for inviting me here.

My name is Willa. I work at Google in product strategy. In particular, I work on two products: Google Earth and Environmental Insights Explorer. And so, what I do is work to bring these products to public sector and city customers and try to understand what their needs are, and communicate that to our engineering teams—and then roll that out and see how it works.

Eggers: Amazing. What are some of the needs and uses that cities, state governments, and federal governments would have of some of your products like Google Earth?

Ng: Sure. I think the No. 1 thing that we hear all the time is data.

Eggers: Yeah.

Ng: With our imagery, with our Street View cars, with our understanding of movement patterns, and then, now, beginning with a lot of these foundational models that our research teams and different sister companies are building.

But we also hear a lot about the ability to analyze and gain insights from the data. And then, in particular, we hear a lot about collaboration and the need to bring people in to the project that you’re working on. And so, that can be: How do we create reports more quickly? How do we tailor reports? How do we share information?

And so, that is going to be a big part of what we start to focus on at Google: understanding the whole journey of how a project gets done, including that sort of last iterative, sometimes painful step of getting buy-in from all of your stakeholders.

Eggers: Right. Well, and you mentioned all the collaboration.

I’ve been doing digital government—e-government—for over two decades now, and I know from citizen developers and community organizations and lots of others who work with cities and the public sector that there’s been so much use of Google’s products and data, and so on. And so, when you’re looking at model complexity and usability and accuracy when building a lot of these geo and sustainability tools for nontechnical users, how do you balance that tension between accuracy and speed and innovation?

Ng: I think the models themselves and the technical backing of it shouldn’t really ever come to mind, unless you’re deeply technical.

I think, for most people, the complexity of the model should be our problem.

I think the only thing that really matters is: Is it reliable? Is it consistent? Is it trustworthy? And I think that is really what we aim for: that baseline level of trust. And then the second thing is, how accessible is it really?

Do you find yourself coming to it, asking questions, asking follow-up questions? Do you find yourself coming back?

Eggers: Right.

Ng: If it’s not easy to use, they won’t come back. And so, I think that those are what we’re really trying to aim for right now—not just for the technical audience, not even for a GIS analyst. I think we’re really trying to get to a more, I would say—I don’t know if this is the word for it—democratized access to this type of technology and analysis.

Eggers: Well, you mentioned trust, and obviously trustworthy AI—a very big issue. And there’s been a lot of polling and surveys around just regular people and if they’re positive or negative on AI and how much they trust it.

You know, right now, the general populace is not necessarily there or is split in some ways. What are some of the ways of building more trust in the data and AI?

Ng: Yeah, and I think that it’s not going to be an immediate thing—and neither should it be.

Eggers: Right.

Ng: If you pull up Gemini today, it’s written right across the bottom: “Gemini can make mistakes. Please double-check.” And I think that should be true for the foreseeable future. Right now, for instance, on Google Earth, we’re investing in geospatial analysis that should look very familiar to many people who work in that field, even though we also launched Earth Gemini, which is a chatbot that can do sort of similar things, in parallel.

And I think that those sort of traditional methods and AI methods will have to run in parallel for a while to gain that trust over time. As you say, I’m doing this analysis here, I trust it because it’s very familiar to me, and I’m doing this analysis simply by asking a question, and those answers are the same or very similar. I think that’s what proves it.

Eggers: So, it’s going to take a little bit of time for people to get a little more used to things and more trustworthy. Cities vary widely in their data maturity, their resources, and even their capacity, institutionally. From your experience, what are the biggest levers that allow a city to make meaningful action once they unlock geospatial sustainability data?

Ng: I’ll come back to that collaboration piece. I think we all sort of know that the geospatial analysis itself is just one small step of the many, many steps that it takes to get to a complete project as a public sector agency—there’s the funding that you have to get, there’s the buy-in that you have to get.

And so, I think that as we’re thinking about AI, you know, it can be used for things like data cleaning, data analysis, interrogating the data, [and] can also be used for things like generating reports and sharing things with your colleagues. So, I really think that those are the things that can be unlocked once you get past the step of using AI for geospatial analysis.

Eggers: Well, and you even mentioned the democratization of it. I did a case study once on a massive slum area in [South] Africa, outside of Johannesburg, where essentially most of the population wasn’t even counted because they weren’t able to do that. And they used geospatial analysis to basically bring those people into the population who are counted, so that they could get additional resources. So, there’s a lot of really, really interesting applications that can actually help a lot of people who are underserved before.

Ng: Yeah, I think there are a lot of lessons that even a company like Google can take from that. I was just at a session this morning where they were using AI and computer-vision detection of dwellings in Africa. And these informal dwellings pop up. They said a quarter of the population was in informal dwellings, and they really had to do a lot of labeling and training of that AI to start recognizing that those are legitimate dwellings. And I think that’s something that we’ll also have to start taking into account as we roll out our geospatial tools and AI tools across the globe: local context is really going to matter. And so, the more feedback that we can get from—and probably the public sector here is the critical player—about what is it that we’re not getting right about your city and your area.

Eggers: Another area, as you mentioned—Gemini. And, by the way, one of the things our son is very excited about is the ability to actually build video games—personalized video games—with AI. I don’t know how fast that is gonna come on board, but I’m sure that’s going to change our households today.

Ng: For the better?

Eggers: We’ll see. We’ll see about that. Generative and agentic AI are being adopted in public transport, like California’s gen AI pilots for highway safety. How do you see technologies changing the daily work of transportation planners and operators?

Ng: Yeah, I mean, like your son, I have done things with AI in the last six months that I never thought I would do. So, it starts off with, AI is going to help me review all of these documents, write a literature review, and then draft me a one-pager. Right?

Eggers: Right.

Ng: Oh, easy. Like, even that, six months ago, was amazing to me. Six weeks ago, I started using AI to help me write SQL queries for the transportation databases that we have. Just again, and again, just a natural language. I need a query that does this. I want to understand, you know, the origin-destination patterns on this day. Bam! Right. It just generates it for you. And then in the last couple of days, I’ve even started [trying] my hand at vibe coding.

Eggers: Oh, you are? You’re doing vibe coding?

Ng: Oh, everyone should give it a try. It’s actually really fun. It makes you feel like the smartest person ever, at least [compared with] the smartest software engineers that you work with. But you start to, without really, you know, I took one coding class in college and did not do well in it, but I can write what a piece of software should do in words, in plain English.

And I can then check to see if it’s doing what I want to do after it writes the code and compiles it, and then creates a little prototype of it. And I think that those are such amazing things. I think that there’s no transportation planner worth their salt who hasn’t imagined and fantasized about sort of the perfect mobility-as-a-service app, right?

Eggers: Right.

Ng: You can kind of build a prototype of that to really show, to a user, what your imagination has come up with, and get feedback on it, iterate on that, and sort of make it better in a way that a simple text description or even a graphical mock could never do.

Eggers: So does this mean—we talked a little bit before the show about education and how this is going to transform, or should transform, education—are we gonna see all these kids who have grown up with Minecraft becoming transportation planners when they’re in their teens? And playing around with all sorts of simulations and scenarios and being able to go into these worlds?

Ng: Oh, I wish, I wish the next step for my five-year-old is transportation planning. I don’t think that that’s what it’s going to be. But I think that that’s one of the really promising parts of AI. We talked about democratizing access—but not just for kids. You work with people in the public sector and quite often the toughest step is making sure that the residents of an area, the citizens of a neighborhood, are fully aware of what’s going on and are participating in crafting the solution together.

And hopefully what AI is able to do is put a lot more of these specialized tools in the hands of people who don’t have that specialized training but can think up a question to ask of a project. And so, I really am hoping that there are more people who can participate really deeply in transportation planning and public sector engagement.

Eggers: And tools that help them to understand all the different trade-offs. If you, you know, put this community here, what the infrastructure needs are going to be, because a lot of this is really about trade-offs.

Ng: Right.

Eggers: And, right now, if you’re a resident and everything, you don’t see all the trade-offs necessarily. Through the ability to democratize the use of things like digital twins, we can do a lot more cocreation with communities and make people feel like they’ve been engaged. And I’m really excited about this all over the world, especially in areas where people have felt like, you know, sometimes the governments weren’t listening to them and that they didn’t have a part in civic life in that way.

Ng: And, to add to that, you know, most of the time, the real deep engagement can only come when someone physically attends a meeting with you. A public meeting with you, sits there, and maybe they move some blocks around a virtual town. But with these new tools, you may start to hear more from people who just can’t make it to that meeting.

And we start to hear more from that silent majority who is benefiting from these tools. And you’re making me flash back to my time in the public sector when we would hold these elaborate public meetings that three people would show up to and they would only be the people who were negatively impacted and really didn’t want the project to happen. Fair enough. But there would be, for instance, bus riders or cyclists who would be benefiting from it, who didn’t have the time to come. How do you then put these very elaborate scenarios and trade-off exercises in front of people?

Eggers: So will we all be avatars in this digital world, interacting in digital and online town halls, but also being able to see, as avatars, all the different decisions and what the trade-offs [are] going to be. How far away do you think we’d be from that?

Ng: I actually would hope not. I actually like seeing people in person. I just think that it would help to expand the pool.

Eggers: Right.

Ng: And of course, and maybe that’s me and my age, where I like to see people in person. But maybe the younger generation will be very, very comfortable with being an avatar and expressing their opinions and wants and needs via something virtual.

Eggers: Yeah. I mean they have done [that], when they’ve opened up a lot of city council hearings, and things like that, they actually have found they have got a lot more young people involved. So, I do like that idea of a hybrid.

Ng: Mm-hmm.

Eggers: The last question I wanted to ask is around this convergence of physical and digital engineering, and we’ve talked a little bit about that. But how can that help transform transportation and a lot of these options, and what are some of the governance challenges of doing that? And you live in New York City—a lot of kind of issues over time with getting everyone to agree on a direction.

California—very similar. Can you talk a little bit about the physical and digital pull-together and how we’re going to handle that from a governance standpoint?

Ng: I think it’s actually a very complex topic.

Eggers: Yeah.

Ng: For one, I am learning a lot of things even at this conference about some of the struggles that cities have with using the technology and AI. There was a gentleman that I spoke with in New York, who is the director of machine learning and AI for the City of New York.

Eggers: Yeah.

Ng: The topics that he brought up were about education of AI and different technologies, both for public sector employees and for the public, informing them that AI is being used and in what ways. And don’t be scared because here are all the details of how it’s being used. It’s upskilling the public sector itself, which I think can be very, very challenging from my time working in the public sector. And then the last challenge that he really talked about was procurement.

Eggers: Oh, yes. Procurement.

Ng: He talked about how the system was set up to buy fire engines and not technology that changes every day. So this need to sort of maybe, especially for a city, change the way that they procure technology is starting to really rise to the top. And I think it really is rising to the top because of these sort of foundationally transformative technologies that can have real benefit, so that the incentive is now there to transform this super burdensome process.

Eggers: Yeah, we are in the middle of working on our annual Government Trends report—and this one’s looking at the future of government. One of the biggest trends we’re seeing all over the world is [within] procurement reform—cities, state level, federal level—and a movement toward what’s almost called “programmable procurement,” essentially, where you’re putting a lot of it into code and then you could move a lot faster.

For the audience, a couple of months from now, you’ll see some of those big changes there. Very excited, actually, on the procurement area because there is so much you can do right now through digitization and using AI.

Well, lastly, any last things you want to mention that you’re doing at Google or that are coming up in the future here?

Ng: I think the most exciting thing—and I’ll just be very selfish about this because my product is Google Earth. We’ve launched vector data sets on top of imagery in Google Earth.

Eggers: What does that mean for the audience? Vector data set …

Ng: Geospatial data sets. I’m sorry. I went a little bit hard left into nerd, and I’m coming back right now. It is information that you can then overlay onto imagery. So, for instance, movement patterns or the locations of trees, the locations of fire hydrants, things like that, on top of imagery. And then, we’ve also recently launched, in October, Earth AI Gemini, which is a chatbot that you can use to start to interrogate some of our imagery data.

Eggers: Well, thank you so much for your time.

Ng: You’re welcome.

Eggers: This is a wonderful conversation, and thank you all for being here.

Eggers: Thanks to Willa Ng from Google for sharing so many great insights. If you’d like more from Government’s Future Frontiers, you’ll find our previous episodes wherever you get your podcasts. And to make sure you don’t miss new episodes, be sure to follow the show on your favorite podcast platform.

This podcast is produced by Deloitte. The views and opinions expressed by podcast speakers and guests are solely their own and do not reflect the opinions of Deloitte. This podcast provides general and educational information only and is not intended to constitute advice or services of any kind. For additional information about Deloitte, go to Deloitte.com/us/about.