On a sweltering August afternoon in Williamsburg, Brooklyn, technologist Chris Whong led a small group of researchers, students, and local community members on an unusual walking tour. We weren't visiting the neighborhood's trendy restaurants or thrift shops. Instead, we were hunting for overlooked public spaces: pocket parks, street plazas, and other spaces that many New Yorkers walk past without even realizing they're open to the public. 

Our map for this expedition was a new app called NYC Public Space. Whong, a former public servant in NYC’s Department of City Planning, built the platform using generative AI tools to write code he didn't know how to write himself – a practice often called “vibe coding.” The result is a searchable dataset and map of roughly 2,800 public spaces across New York City, from massive green spaces like Flushing Meadows–Corona Park to tiny triangular plazas you've probably never noticed.

New York City has no shortage of places to sit, relax, or eat lunch outside. The city’s public realm includes more than 2,000 parks, hundreds of street plazas, playgrounds, and waterfront areas, as well as roughly 600 privately owned public spaces (POPS) created by developers in exchange for zoning benefits. 

What it lacks is an easy way for people to discover these spaces. Some public spaces appear on Google Maps or Apple Maps, but many don't. Even when they do, it's often unclear what amenities they offer and whether they're actually publicly accessible. You might walk by a building in your neighborhood every day but have no idea that it contains a courtyard or indoor plaza open to the public.

The information about where these spaces are exists in government open data, but it’s scattered across multiple agencies and isn’t published in a way that makes it easy for the public to use.

That disconnect—between the wealth of public spaces, information about those spaces, and the difficulty of actually using that information—pushed Whong to try stitching the pieces together himself.

Wall Street Plaza is one of the roughly 600 privately-owned spaces in New York City open to the public. Photo by Tdorante10 | CC BY-SA 4.0.

Assembling the Data Puzzle

The idea for NYC Public Space first came to Whong several years ago while leading a team at the NYC Department of City Planning, building applications to visualize public datasets. While building a map of waterfront public access areas, he saw that there could be value in combining data that already exists about public parks, POPS, plazas, schoolyards, and other public spaces, but is fragmented across different datasets. “I was just like, man, it'd be great to kind of see all these things on one map, as a human trying to use them," Whong says. 

Fast forward to 2024. Whong now works at Mapbox, a mapping software company. In his spare time outside of work, he decided to finally build the unified public space map he'd envisioned years earlier. 

The first step was to gather data on the locations of different public spaces – parks, waterfront access areas, street plazas, POPS, and more. The NYC Parks Department maintains a database of the city’s more than 2,000 parks, the Department of Transportation tracks street plazas, and other agencies document waterfront access points, school playgrounds, and state and federal parkland within city limits. Much of this information was available through NYC’s Open Data portal. Still, it required cleaning and standardization so that each space included the same basic attributes – like its name, location, and type—before they could all be shown together on a single map.

Whong used this dataset to create an initial version of the map. He then refined the dataset by deleting spaces that were included in the public data that didn’t belong on the map. For example, he removed planted medians (which were included in the NYC Parks data) since they do not really function as a public space. Whong also added spaces he noticed were missing from the initial dataset, such as state and national parks.

Vibe Coding the App

Once there was a working version of the map, Whong began to develop the mobile app to put the map of the city’s hidden plazas and parks into people’s pockets.

Whong is an experienced web developer and programmer, but had never built a mobile app. So after reading some tutorials to get started and writing some initial code, he turned to ChatGPT for help. “Using the web interface, you can just copy and paste big chunks of code in it,” Whong explains. “I would take a component…that I was working on, and just paste it in, and say, ‘how would you implement XYZ feature in here?’ … So it was kind of this iterative learning thing.”

While building the app, Whong realized that the map would be more useful if it included written descriptions of the public spaces that could be displayed along with their locations. With thousands of spaces in the dataset, Whong couldn't write all these descriptions by hand, so he used ChatGPT’s application programming interface (API) to generate one or two sentence descriptions for each space. An API allows different software programs to communicate with each other – in this case, allowing Whong's programming scripts to automatically send the public space dataset to ChatGPT in bulk and receive generated descriptions in return. 

The first attempt was a disaster. "I got lucky on the ones that I spot-checked,” Whong says, “...but the bulk effort was not good." The AI returned a lot of incorrect descriptions – for example, claiming that a public space was in a completely different borough. The results improved significantly through several iterations of refining the prompts, including pointing ChatGPT to good sources of information about the public spaces, such as pages on the NYC Parks Department website.

In addition to brief descriptions, the app provides details about the spaces designed to help people understand what they can be used for and what amenities they offer. This includes photos and details such as whether the space is wheelchair accessible, has trees and grass or a view, and whether it has facilities such as restrooms, drinking fountains, seating, or tables. As I’ll explain below, these details are crowdsourced from users who visit the spaces and submit updates through the app. 

The project took a few months of nights and weekends to produce an initial version that shipped around the end of 2024. Whong built the app using Flutter, an open-source software development kit that allowed him to save time by writing one set of code that could be deployed as both an iOS and Android application.

Reality Check: When AI Meets the Street

On that hot August afternoon, the nonprofit Municipal Art Society of New York organized a "Data Day" walking tour to test the NYC Public Space app in action. Our group stopped at half a dozen public spaces, using the app to submit photos and details about their amenities and character. We reflected on how well the AI-generated descriptions compared to what we found on the ground.

At Badame Sessa Triangle, a small fenced green space located at the corner of Withers St. and Leonard St., the AI-generated description captured the essence of the place. It described a "tranquil setting featuring attractive landscaping, benches, and memorial plaques." Standing there, we could see the landscaping, the benches, and the memorial honoring a World War II veteran. The app made it quick and easy to snap photos of the space and add them to the database. 

Image by Dane Gambrell

At Frost Street Plaza, though, we encountered something different. The app described it as a "vibrant plaza" offering a "sunny gathering place for the community to enjoy local boutiques, cafes, and street art." When we arrived, the plaza was primarily empty – just a CitiBike docking station and what looked like remnants of a vendor kiosk that had once operated there. The "vibrant" scene the AI described was nowhere to be found. And, standing in the heat under the Brooklyn-Queens Expressway, I didn’t find the space quite as pleasant as the AI-generated description might’ve suggested. 

Image by Dane Gambrell

This gap between AI-generated text and on-the-ground reality makes sense when considering how the descriptions were created. ChatGPT can pull information from Parks Department websites and other sources, but it can’t know what a space actually feels like on a summer afternoon, whether a vendor kiosk is still operating, or how much traffic noise reaches a plaza.

To close that gap, Whong built a moderation system that leans on crowdsourcing: users can flag inaccuracies, add photos, and submit details about accessibility and amenities that AI and official data can’t capture.

About once a week, Whong reviews these submissions from across the city and approves them for publication, gradually improving the accuracy and richness of the map.

Combining AI, Crowdsourcing, and Open Data to Create Public Value

This story shows the power of combining open data, AI, and crowdsourcing to turn scattered information into a living map of the city’s public spaces: 

• Open data provides the base locations and categories.

• Generative AI offers quick descriptions, even if imperfect.

• Knowledge crowdsourced from the community validates the descriptions and offers additional details that help create a better understanding of what public spaces are really like on the ground. 

Together, the app turns public information into a tool to help New Yorkers answer everyday questions like, “Where can I sit outside near me right now?”

The app also illustrates how AI can act as a force multiplier, helping problem-solvers work faster and do more with less. By leaning on AI for the heavy lifting–generating code and writing initial place descriptions–Whong showed how a single developer can do what might have once taken a small team.

“You still need developers,” he emphasizes. “It’s not that magic where you can snap your fingers and something exists.” 

For public servants, the story shows why open data must be designed for transparency and to fuel civic technology that creates public value.

As Beth Simone Noveck writes in a recent blog post, agencies need to treat open data as civic infrastructure: consistently structured, well-documented, versioned, and maintained as a public good. Investments in data curation and stewardship – the unglamorous work of cleaning, documenting, versioning, and keeping systems maintained – pay dividends when others can build useful tools on top.

Looking Ahead

For now, Whong is keeping the app simple and sustainable. He moderates weekly submissions, fixes bugs as they arise, and doesn't rush to add new features. “If I do nothing on it for the next year other than moderate data, it'll still provide value," Whong says.

That modest ambition might be the most important lesson. Not every civic tech project needs to scale dramatically or become an official government service. Sometimes, filling a specific gap is enough. As one user wrote on Reddit, the app is "such a nice way to find a spot to sit while you’re out and about that isn’t a Starbucks."

"That's probably the best glowing praise you could get for this thing," Whong reflected. The app  gives people an easy way to “see where you are, see what's nearby, and hopefully discover something about the city that you never knew existed, and explore it."

Lead image: View across the East River towards Lower Manhattan from the Brooklyn Heights Promenade by Famartin — via Wikimedia Commons — blurred — licensed under [CC BY-SA 4.0]