A recent study examined the route-planning strategies of London taxi drivers, revealing that they prioritize complex junctions and longer streets during initial mental planning, unlike AI algorithms which use step-by-step methods. This intuitive, efficient approach highlights the unique spatial awareness of expert human navigators and leverages their deep memory of the city's street network. The findings suggest that understanding human planning could improve AI algorithms, particularly in flexible planning and human-AI collaboration. Researchers used response times as a measure of "offline thinking" time, observing that drivers first focus on a global route structure before filling in details. This contrasts sharply with sequential AI planning and demonstrates a more efficient strategy.

https://neurosciencenews.com/ai-taxi-driver-cognition-28380/