AI can now design greener cities, but architects still have the final say

Imagine what Philly could look like if builders turned Vine Street Expressway into a giant park, bringing more fresh air and positive environmental impacts to the city.

That might sound impossible, but AI is already coming up with ideas to make it happen. A group of faculty and students at Thomas Jefferson University’s Institute for Smart and Healthy Cities are experimenting with AI to design green infrastructure in Philadelphia just like that. The institute worked on a project this year developing ideas for “urban interventions” that mediate climate change using AI tools, Edgar Stach, director of the institute and architecture professor, told Technical.ly.

“We tailored the design process based on your location and the need to improve environmental aspects in that location,” Stach said.

The researcher learned that not only is generative AI a great tool to come up with ideas, but it also helps to see the impact of designs on the surrounding environment. The Institute for Smart and Healthy Cities researches the way cities are built and how it connects to the environment, housing, transit, public infrastructure and health.

AI image generators and analytics tell builders what’s sustainable

Students used text-to-image AI programs and image-to-image generators to come up with ideas and tweak their results, said architecture student Martin Merroth. The prompts asked AI programs to develop solutions to problems like congestion, traffic, pollution and walkability in Philadelphia.

It took a wide variety of AI tools for this project, including Midjourney, LookX AI and Microsoft and Adobe’s AI generators which all helped the students create new images. They also used analytical tools like Forma.ai to look at site studies, wind direction studies, noise studies and sound pollution to give further insights on how a build impacts the area.

The students input prompts, and the AI programs spit out image concepts based on generally agreed upon concepts for better urban living such as adding parks, green infrastructure and pedestrian walkways, Merroth said.

The project team picked major sites in Philly such as Broad Street, the Vine Street Expressway and Schuylkill Yards. The students decided which places to target based on what they thought could benefit from greener infrastructure, meaning the buildings that try to solve urban and climate challenges.

For example, the institute asked the AI programs to come up with a “greener” transportation hub on Broad Street. Instead of a bulky steel structure, AI presented two options that prioritize light, greenery and blending into the surrounding neighborhood.

“We want to increase the quality of life in the city,” Stach said. “The prompts needed to address this. It’s not just a beautiful rendering. It has a meaning behind the rendering.”

Professional architects still have the final say

Despite the advancements brought in by AI, much of the traditional architecture process still remains the same. In the future, Stach sees AI being used in more analytical tools so that architects can see the environmental and societal impacts of their designs.

For now, though, Stach and his colleagues use AI as a concept generator before moving into the traditional 3D model making process. AI mainly helps architects come up with a lot of ideas quickly, according to Stach. A professional architect still needs to look at the designs and edit them, but using AI fast tracks the front end of the process, he said.

“It’s extremely powerful to communicate your initial ideas, your concepts,” Stach said. “Once you have that concept, you can put it on a test stand and run some analytics [about] how that urban intervention would actually really work.”

Sarah Huffman is a 2022-2024 corps member for Report for America, an initiative of The Groundtruth Project that pairs young journalists with local newsrooms. This position is supported by the Lenfest Institute for Journalism.