"On 5 November 2014, Lufthansa Flight 1829, an Airbus A321 was flying from Bilbao to Munich when the aircraft, while on autopilot, lowered the nose into a descent reaching 4000 fpm. The uncommanded pitch-down was caused by two angle of attack sensors that were jammed in their positions, causing the fly by wire protection to believe the aircraft entered a stall while it climbed through FL310. The Alpha Protection activated, forcing the aircraft to pitch down, which could not be corrected even by full stick input. The crew disconnected the related Air Data Units and were able to recover the aircraft."
The computer thinks the plane in a otherwise normal climb (attitude, airspeed, power, rate of climb) is stalling? That's just...we cannot call this kind of automation anything like a pilot. The term "autopilot" is too much anthropomorphizing. A pilot reacting this way in this same situation with the same information, they would be incompetent or crazy to trust two AOA sensors and assume a stall. If a crash ensued they absolutely would be blamed for it. Not the goddamn sensors.
And calling such a thing a safeguard when it can fail danger like this? A 4000fpm decent as a correction for a stall in a normal climb? That's just batshit. I'm now in a dive recovery operation, itself with higher risks than a stall in the midst of a climb! A stall in a climb is not more dangerous than a 4000fpm dive. Think of throwing a football, it's stalling the entire way, all that's going to happen is the plane stops climbing as fast as it was and the nose is supposed to come down (that's why planes must have positive static stability) and viola we're flying again!
Let's pretend it is true you are somehow stalling in a normal climb (an instantaneous increase in angle of attack): what do does a pilot need to do? Neutralize controls, that's it. Maybe the attitude goes somewhere between 5 and 0 degrees. But a 4000fpm decent? pUke.
(I'm a pilot and former CFII, so I feel sufficiently entitled with my biases to comment with such editorialization.)
The only reason we humans tolerate such unsophisticated systems arriving at weird and dangerous edge cases is because the automation has in fact helped to make flying so much safer. But so have many other things: components are so much more reliable, turbine engines run for many thousands of hours between overhauls, they're really incredible. The training is a lot better. CRM has made things way safer! It's not just the automation. And we really shouldn't be cutting automation so much slack either. I mean for fucks sake, 4000fpm? Crazy. But it's an edge case and it should not be accepted with "oh well the pilots recovered! AOK!"
Obviously if the pilots had not recovered a central theme would be "the pilots didn't recover! INCOMPETENT PILOTS! THEY DIDN'T FOLLOW THE CHECKLIST! WARGARBLE!
Automation is not anything like a human. So accusing it of being crazy, or trying to murder people isn't exactly accurate. But I think it's reasonable to hold it accountable to things that look like "insane" behavior when it fails this spectacularly. Its just a sequence of narrowly defined laws, with each law doing its specific thing, without any regard to other information. Very myopic worldview. And then having more authority to take (wrong) action than pilots have to correct it?! What a bad sequence!
I used to be one of the "automate all the things!" crowd, but as I've been exposed to more and more fields, I've come to realize that automation's place is to augment and assist, not replace. There are diminishing returns to increased automation, as at the end of the day, the user of the automation needs to be 10% smarter than the piece of equipment, which quickly becomes a more and more challenging task the more the challenge of overseeing the automation takes away from doing what you are there to do in the first place. In this case, flying the plane.
