Researchers have studied the thought processes of London taxi drivers, who are renowned for their extensive knowledge of over 26,000 city streets, as part of an investigation into how artificial intelligence can improve route-mapping techniques.
Researchers have studied the thought processes of London taxi drivers, who are renowned for their extensive knowledge of over 26,000 city streets, as part of an investigation into how artificial intelligence can improve route-mapping techniques.
Unlike a traditional GPS device that explores all potential routes to find the best one, researchers at the University of York, in collaboration with University College London and the Champalimaud Foundation, discovered that London taxi drivers approach route planning with a smarter strategy. They focus on the most challenging points first and then build the rest of the route around these difficult areas.
Applying current computational models to analyze human planning can be tough in real-world scenarios and on a larger scale. Therefore, the researchers observed the decision-making time of London taxi drivers as they planned trips to various locations within the city.
Past research has highlighted the unique nature of a London taxi driver’s brain, noting that they possess a larger posterior hippocampus than typical individuals. This brain region changes in size due to their extensive driving experience.
Dr. Pablo Fernandez Velasco, a British Academy Postdoctoral Fellow at the University of York, commented: “London is extremely complicated, making the ability to plan a car journey ‘on the fly’ a significant accomplishment.
“If taxi drivers were to plan their routes in a linear fashion—street by street, as most people do—we would observe significant variations in their response times depending on their position along the route.
“In contrast, they consider the entire street network, concentrating on the most crucial junctions first while applying theoretical metrics to identify what matters. This method is exceptionally efficient, and it’s the first time we’ve been able to analyze this process in real-time.”
The findings reveal that taxi drivers utilize their cognitive abilities far more effectively than existing technology, suggesting that insights from expert human planners could enhance AI advancements in numerous ways.
Dan McNamee from the Champalimaud Foundation stated: “Developing future AI navigation systems could greatly benefit from the adaptable planning techniques employed by humans, especially when there are numerous environmental factors and variables to consider.
“Another way to improve these technologies is by incorporating knowledge from human experts into AI algorithms meant to work alongside people. This is crucial because, for an AI algorithm to optimize its interaction with a human, it must understand human thought processes.”
Professor Hugo Spiers from University College London added: “This research definitely supports findings from other studies—that the London taxi driver’s brain is remarkably efficient and its larger volume is effectively utilized to navigate a city as complex as London.”
The research received support from the British Academy, the EPSRC UK, and Ordnance Survey and was published in the journal PNAS.
