Uber wants to know how drunk its passengers are

The ride-hailing service may be looking at ways to tailor its app depending on how much a user has had to drink.

A new patent application suggests Uber is looking to use machine learning to gauge the sobriety of its users.

The application uses fairly broad terminology, but specifies the aim of predicting “uncharacteristic user states” through the analysis of customer data. 

By using a phone’s gyrometers and location data, Uber could hypothetically track how fast you were moving or how upright you were standing, while a measure of a user’s “data input accuracy” could draw conclusions from how hard your smartphone’s autocorrect was working.

This information would allow Uber to further personalise its service, ensuring pick-up points were in well-lit areas, or even presenting a simpler version of the map interface on its app.

No YOU'RE drunk

Not everyone is likely to welcome the prospect of Uber harvesting yet more of its customer’s data, however. Uber has come under fire before for its misuse of customer data, which included Uber employees being able to track the location of ex-partners over a centralised ‘God View’ tool.

While Uber is attempting to rehabilitate its public image, users may be wary of an algorithm that can tell how many cocktails they’ve downed on a night out. And while the data could help drivers cater to user needs – maybe with a pre-reclined seat, or a helpfully placed bucket – it could also pave the way for further breaches of trust.

Henry St Leger

Henry is a freelance technology journalist, and former News & Features Editor for TechRadar, where he specialized in home entertainment gadgets such as TVs, projectors, soundbars, and smart speakers. Other bylines include Edge, T3, iMore, GamesRadar, NBC News, Healthline, and The Times.