Facial recognition is not ready to spot wanted motorists

If you’re the kind of person that feels uneasy about governments’ enthusiastic adoption of facial recognition software for law enforcement, then you’ll be pleased to hear that it’s nowhere near ready to catch wanted motorists.

The Wall Street Journal has obtained an email from the Metropolitan Transportation Authority (MTA) revealing the results of a 2018 test on the Robert F. Kennedy Bridge in New York, and the results aren’t pretty. The software failed to spot a single face “within acceptable parameters,” which isn’t ideal for such a high-traffic area.

According to an MTA spokesperson, the trials will continue on the Robert F. Kennedy Bridge, as well as in other locations. The good news is that the only way is up, I guess.

Related: Best dash cams

In lab conditions, facial recognition of motorists works pretty well. Oak Ridge National Laboratory managed over 80% accuracy when identifying faces behind the steering wheel, but crucially these tests were at lower speeds than actual motorists hit when crossing the East River.

Maybe that just means this is a poor choice of location for cameras, and the police should target areas of gridlock. All the same, not being able to identify wanted criminals going at speed does sound like something of a flaw in the system when you’d expect them to be in more of a hurry than your average commuter.

That may be for the best. Setting aside very legitimate privacy concerns about the rights of innocent citizens not to be tracked when going about their business, facial recognition is currently riddled with problems. Just last year, Google’s head of cloud computing stated that the system carries “inherent biases,” which isn’t ideal for something governments seem keen to include in future plans for law enforcement.

Do you see the adoption of facial recognition as essential for public safety or a violation of personal privacy? Let us know what you think on Twitter: @TrustedReviews.