The NTSB released their much-anticipated findings on the Uber self-driving crash last week. Ars Technica called the report “a damning picture of Uber’s self-driving technology.”

Per AT:

“The report confirms that the sensors on the vehicle worked as expected, spotting pedestrian Elaine Herzberg about six seconds prior to impact, which should have given it enough time to stop given the car’s 43mph speed.”

“The problem was that Uber’s software became confused, according to the NTSB. ‘As the vehicle and pedestrian paths converged, the self-driving system software classified the pedestrian as an unknown object, as a vehicle, and then as a bicycle with varying expectations of future travel path,’ the report says.”

“Things got worse from there.”

“‘At 1.3 seconds before impact, the self-driving system determined that an emergency braking maneuver was needed to mitigate a collision. According to Uber, emergency braking maneuvers are not enabled while the vehicle is under computer control, to reduce the potential for erratic vehicle behavior. The vehicle operator is relied on to intervene and take action. The system is not designed to alert the operator [says the NTSB report.]”

“The vehicle was a modified Volvo XC90 SUV. That vehicle comes with emergency braking capabilities, but Uber automatically disabled these capabilities while its software was active.”

“The vehicle operator—who had been looking down in the seconds before the crash—finally took action less than a second before the fatal crash. She grabbed the steering wheel shortly before the car struck Herzberg, then slammed on the brakes a fraction of a second after the impact. The vehicle was traveling at 39mph at the time of the crash.”

“Dashcam footage of the driver looking down at her lap has prompted a lot of speculation that she was looking at a smartphone. But the driver told the NTSB that she was actually looking down at a touchscreen that was used to monitor the self-driving car software.”

“‘The operator is responsible for monitoring diagnostic messages that appear on an interface in the center stack of the vehicle dash and tagging events of interest for subsequent review,’ the report said.”

Citylab’s report on the findings highlights the “handoff problem” we may be looking at with level 3 automation by highlighting a nearly 25-year old NTSB report on airline crashes:

“Over the period of that study, aviation had moved into the automation era, as Maria Konnikova reported for The New Yorker in 2014. Cockpit controls that once required constant vigilance now maintained themselves, only asking for human intervention on an as-needed basis. The idea was reduce the margin of error via the precision of machines, and that was the effect, in some respects. But as planes increasingly flew themselves, pilots became more complacent. The computers had introduced a new problem: the hazardous expectation that a human operator can take control of an automated machine in the moments before disaster when their attention isn’t otherwise much required.”

Citylab goes on to note that the so-called “safety drivers” in Uber’s self-driving cab fleet often grew complacent in the same way. One safety driver is quoted as saying “When nothing crazy happens with the car for months, it’s hard not to get used to it and to stay 100-percent vigilant.”

So despite the debate over whether Uber’s AI was set to ignore too many false positives or whether it used too few sensors of one type or another or whatever the argument is, it looks like there’s precedent here: it’s something the airline industry has been through and learned from. Perhaps it’s time to start paying attention to how the aviation industry has learned how to all but eliminate fatal accidents — if the burgeoning AV ride-hailing industry can’t pull it off, we may be stuck waiting for level 4-5 technology for self-driving taxis.