Uber is turning to artificial intelligence (AI) to know when a rider has had too much drink.
The ride-sharing giant has published a patent application to identify drunk passengers.
It describes an algorithm that will weigh a variety of factors, like typos, how precisely a user clicks, walking speed and the time of day.
For example, a rider clumsily typing at 1 a.m. on a Saturday would likely be considered not sober.
Uber plans to use that technology to better tailor its rides, either by warning drivers or matching them with drivers with relevant experience or training.
The system, described by members of Uber’s Trust & Safety team in 2016 and published yesterday, tracks how someone typically uses Uber’s app: how quickly they type (and with how many typos), how precisely they click on buttons, their walking speed, and the way their phone is typically held or dropped on any given day.
How drunken or disorderly the AI assumes a rider to be is ultimately what shapes what their ride looks like, according to the patent application.
A driver set to pick up a particularly tipsy passenger might get a notification on their phone beforehand, or they could be told to meet the rider at the designated location a bit later to give the person more time to get to the car before driving off.
Depending on their sobriety levels (or lack thereof), a passenger may only match with drivers who are trained or experienced with handling people in this state, or the rider might be prevented from joining a carpool.
Read also: iPhone X gold images revealed in a leak by FCC
Another disturbing possibility comes from the drivers who would not use this indicator as a deterrent but as a draw. There have been more than 100 cases of assault and abuse at the hands of Uber drivers over the past four years. A common thread through many cases, according to court documents obtained by CNN, was that the victims were either heavily inebriated — if not fully passed out — while the assault took place. Uber’s planned AI feature could, in theory, let drivers easily scout neighbourhoods to find drunken passengers for this purpose.
Uber declined to comment on these issues.
