Here’s why Google had to scrap handover tech on its self-driving cars

Google has had to change its vision for self-driving cars, and it’s because of humans being useless again. According to a new report on Reuters, Waymo, Google’s self-driving operation, has scrapped the idea of handing over control to drivers, because tests showed most simply weren’t paying attention to the road.

Rather than a fully-autonomous solution, Waymo had been testing driverless cars that handover to the driver when they encounter tricky situations, but Waymo says its test drivers were pretty awful. Instead of being alert, and ready to take control in situations deemed too tricky for AI, the human test pilots were being too human.

According to the new report, test pilots were “napping, putting on makeup and fiddling with their phones as the vehicles travelled up to 56 mph.”

“What we found was pretty scary,” John Krafcik, the head of Waymo, told Reuters. “It’s hard to take over because they have lost contextual awareness.”

The tests were actually conducted around four years ago in 2013, but they were first shown in public earlier this week. Waymo said the system used an alert sound that prompted used to take control – but that it simply wasn’t safe.

All or nothing

These tests are disappointing and interesting at the same time, and they have a huge bearing on how autonomous cars will first appear on our roads. Right now, the semi-autonomous technology you turn on – for example, on the motorway – seems to be safe. Sadly, the idea of a system that’s on until it finds something tricky, doesn’t seem to be safe at all.

It’s important to stress that these tests were done using Google’s own employees, and if they were distracted, it’s horrific to imagine how lax non-test subjects would be.

Instead, it looks like if we bring autonomous tech to the road, we’ll need cars that are fully driverless, and able to handle all real-life situations. If Waymo’s tests prove anything, it’s that right now human drivers simply can’t be trusted.