Cities from Madrid to Malaysia are using AI for traffic, planning and resident questions. Here is what is working and what is still unproven.
Cities around the world are putting AI to work on everyday jobs: timing traffic lights, answering resident questions and testing street plans before anything is built. NVIDIA, the chipmaker, laid out a batch of examples on its blog ahead of the Smart City Expo World Congress in Barcelona, running Nov. 3-5. The most striking number comes from Malaysia, where traffic tests cut peak-hour travel times on some corridors by up to 26%.
Cities collect huge amounts of data from cameras, sensors and city systems, but often struggle to turn it into action. The new approach uses AI agents, which are programs that can answer questions and take steps on their own, along with digital twins, which are computer copies of a place used to test changes safely. NVIDIA says city teams may not need a big investment to start. Depending on the job, a workstation or a small GPU system, a computer built for AI work, may be enough.
In Malaysia, ITMAX uses an open source traffic simulator to test signal timing, which is how it reached the 26% figure. In Spain, the agents handle municipal procedures, appointment requests, urban-planning incidents and general questions. In Taiwan, Linker Vision and the Dutch architecture firm MVRDV model how a city forest would grow. Design rounds that took two weeks now take less than an hour.
Residents may see more chat assistants at city offices, and in places like airports and hospitals, where LiveX in the San Francisco Bay Area is building conversational AI hosts. Businesses that sell to cities, or rely on permits and local services, may see faster responses and more AI in the process. Cities can also share computing power across departments and neighboring towns instead of each buying its own.
Nearly all of this comes from NVIDIA and its partners, and the post gives no independent checks. The 26% result is an upper limit, not an average. Costs are not listed. Several projects, such as the Tanger Med port work in Morocco and the camera rollout in Malaysia, are described as planned or still scaling. The post says local governments stay in control of how sensitive data is used, but it does not explain how.
NVIDIA Vice President of EMEA Paolo Guglielmini will give a keynote at the Barcelona event, followed by a panel. Expect more city announcements there. Whether these tools deliver the same results outside company-run tests is the open question.
City AI is moving from pilots to real use, but the big results so far are company-reported and often still planned.
World models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasksA University of Chicago law professor asked Claude to fix page numbers. It deleted his syllabus and a year of notes instead. Dropbox saved him, not the AI.
Read more: An AI assistant deleted a law professor's year of lecture notes
