Uber and Lyft are seen as the cure to drunk driving. But that forgets a pretty important point: There’s too drunk to drive, but sober enough to keep it together. And then there’s too drunk to be in a car with anybody, whether that means an inability to keep down the tequila slammers and spicy wings from the night’s festivities or picking a fight with anybody in range. Uber drivers aren’t janitors or police, after all, and there’s no amount of money to make some riders worth it. So how do you sort the sober from the hammered? Uber is turning to AI.
Metro is reporting on a patent that Uber has filed that lets it use past data, and current behavior, to figure out if a rider is so hammered they might be a risk to the driver, or as Uber calls it in the patent, “A travel coordination system [that] identifies uncharacteristic user activity and may take an action to reduce undesired consequences of uncharacteristic user states. The system uses a computer model to identify user and trip characteristics indicative of the unusual user state.”
In other words, if you’ve gotten so drunk an Uber driver has to shovel you out of the car, the app will do more than just let the driver ding your rating — the app will put the time, location, and other data around the ride you hailed on file and refer to that data in the future.
There are some pretty glaring problems with this, the most basic being that while Uber’s exact impact on drunk driving is an open question, a ride-hailing service being available is at least a tool to help. Also, it assumes drunks repeat patterns. It’d be a lot better if Uber simply allowed a driver to decline the ride if somebody is too drunk, without a penalty. But then again, it probably wouldn’t kill somebody who regularly barfs in the back of cars to ask themselves if they need to make some changes in their life choices.