A call for clarity on self-driving terminology

Driverless or self-driving — autonomous or automated — automation or autonomy — these terms are often used interchangeably. That's regrettable because they don't all mean the same thing.

Automation, for example, describes the presence of automatic equipment that's used as one part in an overall process. Autonomy, on the other hand, confers the idea that a system is governing itself and all actions. Self-driving implies that a vehicle is being driven without a human involved, while driverless might indicate no one or nothing is in control at all.

Those terms are just the start. Add the likes of semi-autonomous, partially self-driving and driver-assist feature, and the language landscape becomes more cluttered. Throw in industry jargon such as "Level 2 Plus" and Elon Musk's special dictionary entries for "Autopilot," "feature complete" and "fully self driving," and this confounding mix is complete.

It's time to clean up the mess.

Using accurate language can be a matter of life or death. We've already seen multiple fatal crashes involving the use of Tesla's Autopilot, in which motorists treated a driver-assist feature as a self-driving system and let their attention and oversight wander from the road.

Everyone from industry executives to journalists to car salespeople has a responsibility to get this right.

Start with the simple fixes. Banish "partially self-driving" and "semi-autonomous" from our collective vocabulary. These are the vehicular equivalent of being a little bit pregnant. Either the system maintains responsibility for the driving operations or a human holds that responsibility. There can be no in between.

Next, take a good look at the SAE Levels of Automation. Nowhere is there a definition for the "Level 2 Plus" notion that has become so commonplace across the industry over the past 18 months. It's an invention of those who want to convey the progress and sophistication of their driver-assist systems without actually accepting the responsibility that comes with higher levels of automation.

Using the SAE levels themselves brings limitations. They make sense for insider industry discussions, but there's a real disconnect in attempting to use them as a springboard for education of broader consumer audiences. And NHTSA has done no favors with its "highly automated vehicle" grouping of Levels 3, 4 and 5.

I like the remedy recently proposed by Waymo CEO John Krafcik.

"If every time someone talked about a 'Level 2 system,' they just said, 'driver-assist system,' then it'd be so much easier," he said. "Maybe the definitional framework is wrong. ... If you need a driver's license, then you shouldn't call it fully self-driving. If you need a driver's license, call it a driver-assist."

That seems like the right way to draw a firm line between driver-assist and self-driving technology.

If a human driver's license is needed for legal vehicle operations, if a human has responsibility for either control or oversight of a system, if a human must be available to accept a handoff from a system, then it's a driver-assist feature. On the other hand, if a vehicle can complete a journey from point A to point B without any role for humans, then it's afforded the terminology of self-driving system.

Most importantly, we need to guard against conflating the two.

The language landscape is crowded and jargon-riddled right now. It may only get worse as companies add their brand-specific names to the mix. But there is good news: We're still in the early days. There's still time to get this right.