Uber is looking at the idea of using artificial intelligence (A.I.) to determine if riders have had one beer too many, and then tailor its service accordingly.
A patent application filed by the ridesharing company this week details how A.I. could be used to detect small changes in a rider’s behavior as they use the app to request an Uber.
Titled “predicting user state using machine learning,” the patent doesn’t explicitly mention drunkenness or drug use. But its contents, which speak of “uncharacteristic user states” and “identifying a normal or abnormal state of the respective user,” suggest such matters may be at the center of the idea. Other conditions that could also conceivably lead to uncharacteristic behavior by a rider could also include extreme tiredness.
So how would it work? According to the filing, A.I. smarts built into the app would monitor behavior such as typing speed and accuracy, as well as walking speed and direction, before interpreting the data to determine whether the rider is compos mentis or utterly mangled.
In other words, if you’re typing slower than usual, making more mistakes than you ordinarily do, and perhaps dropping the phone or staggering all over the place, the software is likely to conclude that you’re intoxicated.
The algorithm could also make use of time and location data, taking special note of whether or not the ride request is coming from an entertainment area with bars and clubs at the end of the night.
If the system concludes that the rider is inebriated, it could deal with the Uber request in a number of ways. For example, it might match the razzled rider with a more experienced driver, or one trained to handle such potentially disorderly passengers. It might also prevent drunk riders from taking a pooled ride. At the very least, it would serve as notice to the driver that the rider they’re about to collect may be the worse for wear.
Of course, if the algorithm detects someone in a particularly bad state, a driver may simply refuse to pick them up for fear of them vomiting in their vehicle, or worse.
Some Uber riders, however, may find the idea expressed in Uber’s patent somewhat troubling, especially as it could offer a way for predatory drivers to target vulnerable riders who may not be in full control of their faculties.
Privacy advocates, too, may not like the sound of it. Uber doesn’t have the best track record when it comes to handling customer data, and holding information on when its riders are perceived to be sozzled or sober may leave some riders feeling uncomfortable.
When asked about the patent, Uber boss Dara Khosrowshahi told the NY Post that “it’s kind of a cool idea,” adding that if it made Uber safer, “it’s certainly something we would look at.”
It should be pointed out, however, that at this stage it’s simply an idea outlined in a document, so it may never see the light of day. We’ll keep you posted.
- A literal minority report: Examining the algorithmic bias of predictive policing
- Uber vs. Lyft
- The best mileage apps for small businesses in 2020
- Autonomous ridesharing isn’t dead: How Waymo is adapting to the post-COVID era
- ‘No mask, no ride’: Uber extends face-covering requirement indefinitely