Google Research has demonstrated that AI-driven navigation adjustments can significantly reduce urban traffic congestion and emissions. A six-month experiment across 10 major US cities modified the Google Maps algorithm to subtly reroute a small fraction of trips—under 2%—away from pre-selected congested road segments. This intervention, applied systematically, aimed to disperse traffic rather than optimize individual routes.

The study, published in Nature Cities, revealed that these targeted rerouting recommendations resulted in a median increase of approximately 2% in driving speeds on targeted segments. This led to a corresponding median decrease of 0.5% to 1.0% in fuel consumption. Even segments receiving redirected traffic saw improved speeds, particularly during peak hours. These improvements translate to potentially thousands of tons of CO2e emissions saved annually per city.

This research establishes a framework for system-wide traffic optimization, moving beyond individual trip planning. By coordinating a small percentage of journeys, the system achieved network-wide benefits for all road users, including those not using the navigation app. The findings suggest that networked navigation technology can proactively shape traffic flow, offering a blueprint for future smart city initiatives involving dynamic control and real-time network optimization.