Taxi companies could dispatch cars more efficiently, give their drivers more business, and make customers wait less.
The Japanese wireless carrier worked with a leading taxi company to train the AI, teaching it how to predict the size and location of demand based on actual ridership data from 4,425 cabs, coupled with information about such factors as weather, and data about people's movements based on the locations of their mobile phones.
The AI ultimately learned to predict localized taxi usage demand 30 minutes out for squares 500 meters on a side in a grid. It can forecast demand for rides with 20% or less error more than 90% of the time, according to Docomo.
For the field trials, 12 drivers used tablets updated every 10 minutes with ridership predictions to decide where to go to find waiting customers. Daily business rose 20% thanks to the information, a participant said.
Docomo said it is thinking to commercialize the technology as early as the second half of fiscal 2017.
This is an example of how AI is starting to find applications in everyday life, beyond specialized areas like drug discovery.
Another is a system that can predict how late trains will come after accidents occur. Fujitsu plans to market it starting in the first half of fiscal 2017 to providers of transportation information services.
If you want to read this article in Japanese, please see the following link: