Gaby
According to Chinese researchers, AI can help urban planners realize their visions for land development more quickly. Their study demonstrates how machine learning can create more efficient spatial designs than humans, all while using significantly less time.
AI Plans Developments More Efficiently and Quickly Than Humans
Artificial intelligence has brought to life the concept of urban planning as a cool green city filled with parks, pedestrian and bike paths, and buses that transport people to shops, schools, and service centers. The work took mere minutes to lay out this idea, where everything is planned so that essential needs and services are within a 15-minute walk, promoting better public health and reducing vehicle emissions.
Automation expert Yu Zheng and his colleagues sought new solutions to improve city development, which are rapidly becoming overcrowded and concrete jungles. They developed an AI system to tackle the most exhausting computational tasks in urban planning. It turned out that artificial intelligence creates urban plans that outperform human designs by about 50% across three metrics, including access to services and green spaces, as well as traffic levels.
AI Saves Significant Time in City Planning
Starting small, Mr. Zheng and his team tasked their AI model with designing urban areas just a few square meters in size (approximately 3 by 3 blocks). After two days of training and utilizing several neural networks, the AI system found the ideal layout for roads and land use to align with the 15-minute city concept and local planning needs.
While the model developed by the researchers has some features to expand its use for planning larger urban areas, designing an entire city would be much more complex. Scientists estimate that developing a neighborhood consisting of 4 by 4 blocks involves twice as many planning decisions as a 3 by 3 block area.
However, automating even a few steps in the planning process can save a considerable amount of time. The AI model computes certain tasks in mere seconds, tasks that would take human planners between 50 to 100 minutes to complete. According to the researchers in an interview with Sciencealert, automating the most labor-intensive tasks in urban planning will free planners to focus on more complex or human-centered tasks, such as community engagement and aesthetics.
Not Full Automation, But Collaboration Between AI and Humans
Mr. Zheng’s team believes that AI will not replace humans but will serve as an assistant to urban planners. It will be able to create concept designs optimized by algorithms, which will then be reviewed, adjusted, and evaluated by human experts based on community feedback.
This final step is crucial for good design, as noted by research fellow Paolo Santi. He states that urban planning is not just about allocating space for buildings, parks, and functions; it’s about designing a place where urban communities will live, work, interact, and thrive over the long term.
By comparing the workflow of humans and AI with designs created solely by humans, Mr. Zheng and his colleagues found that the collaborative process could increase access to essential services and parks by 12% and 5%, respectively. The researchers also surveyed 100 urban designers who were unaware of whether the development plans were created by humans or artificial intelligence.
In this context, AI received significantly more votes for some of its spatial designs, while for other plans, survey participants did not express clear preferences. However, as the researchers point out, the real test will undoubtedly be in the communities built based on these plans, measured by reductions in noise, heat, and pollution, as well as improvements in public health that better urban planning promises to deliver.
