Philbrick: The challenge of implementing AI in transportation

San Jose sits at the center of innovation, including when it comes to the use of artificial intelligence and transportation. Headlines frequently espouse the latest development in smart tools, but headlines are the easy part. The real challenge is making those tools work in the messy, real-world conditions of mobility systems.

AI has been shown to have enormous potential to improve mobility, but even though the technology exists we have yet to see it used at scale. We don’t have an innovation problem — we have an implementation problem.

The foundation for AI integration into transportation systems already exists. AI needs vast amounts of data, and the data is there: Transportation agencies collect information from traffic signals and sensors, GPS and fleet data, transit ridership patterns, incident reports, rider feedback and more.

How do we use it? In the words of researcher Adam Cohen, “Within transportation, AI applications are diverse, including automated driving, traffic management, predictive maintenance and cybersecurity. Use cases can also encompass optimizing traffic flow through adaptive signal control, enhancing safety via real-time incident detection and improving public transit efficiency with dynamic routing.”

For instance, studies have shown adaptive signal control improves average performance metrics like travel time and fuel consumption by 10% or more.

Last year, a CSU research project done in collaboration with Caltrans developed and tested an algorithm to identify, classify and track different equipment and workers on construction sites — which can make data collection for maintenance and safety much faster. Another ongoing project aims to address rush hour congestion and emissions by upgrading the street traffic signal through an AI algorithm that would let the traffic signal system dynamically adjust to changing traffic patterns, thus reducing wait times and environmental impact.

With both dense urban corridors and sprawling suburban networks, San Jose reflects the broader challenges facing transportation systems nationwide. The city has significant potential to be a testbed where these kinds of extraordinary innovations can be tried in complex, real-world scenarios. So why isn’t it already?

It is increasingly, to some degree. In February, San Jose announced the city’s public buses are operating 20% faster because of AI implementation. But we still see many pilot projects — think autonomous vehicle pilots that garnered national attention — that never scale regionwide, statewide or nationwide because of funding constraints, workforce capacity and siloed systems with interoperability issues.

A successful pilot is only the first step. If the innovation finds success in the lab, but remains there perpetually, what’s the point? We need to address this gap so that we not only have successful pilots, but successful deployments.

What needs to change?

First, funding must be structured to support full deployment, not just pilots. That means investing in system integration, long-term operations and workforce capacity, not just the excitement of the initial testing.

Second, agencies must prioritize breaking down data silos. That requires shared standards, interoperable systems and coordination across transit, roadway and freight networks — not as separate systems, but as one connected mobility ecosystem. Transit, roads and freight rail are all part of one greater system for mobility, and if we look at only one piece of the puzzle, we can never see how the systems fit together and — in the big picture — affect how people live and move.

Finally, while AI has the potential to improve mobility, including safety, it also poses risks. However, these are not reasons to delay implementation. They are reasons to implement thoughtfully, with intentional safeguards built in from the start.

The technology is ready. The data is there. What’s needed now is the commitment to bring these solutions into the systems we all rely on every day.

San José Spotlight columnist Karen E. Philbrick is the executive director of the Mineta Transportation Institute, a research institute focusing on multimodal surface transportation policy and management issues. Her columns appear on the first Thursday of every other month.