A reader writes:
This has been an issue in the civil aviation industry for decades. There's an old joke, "Future planes will have two pilots: a human and a Doberman. The dog's job is to bite the pilot if he tries to touch anything."
And to some extent that is becoming true. Every airline has standards about how much of a given flight can be flown by the pilot in command; some allow for more "hand flying" than others. But mostly, aside from some takeoffs and most Category 3 (virtually zero visibility) landings, the automation does most of the work. The crew just sets the avionics to comply with a preset flight path, and modifies or hand flies any changes if required by air traffic controllers.
But the flight crew's job is still essential to manage it. Someone needs to be at the helm if the computers get it wrong, and sometimes they do. ATP-rated flight crews are tortured every six months in simulator sessions where the instructor sits behind a console and throws multiple failures at them intending to overwhelm the pilots. Most of the notable aviation disasters were caused my multiple equipment failures.
Can you really develop software that can deal with things it isn't wasn't programmed to? Not yet. But the human mind can, thanks to intuition, experience, recurrent training for worst case scenarios, and as you alluded to, the first person to arrive at a crash site is the flight crew.
As for Philip Ross's article, he has something of a clue. But in a touchy situation, like a Cat 3 landing, where the only criteria for the pilots is to make sure that they can see a runway 'environment' before the plane touches down. Their hands are hovering over the flight controls and computer disable switch, in case it gets it wrong. It is FAR from a passive experience for the pilots.
To quote Ross, "Even Captain Chesley "Sully" Sullenberger, the pilot who deftly ditched his Airbus 320 airliner in New York City's Hudson River after its jet engines swallowed some geese, owed much to his onboard software—among other things, it managed the plane's angle of descent so as to avoid a stall." Great, the avionics had electrical power so wisely, he used it, to ease his enormous load. But what if the avionics failed?
As an aside, the LGA to Hudson River scenario is a global standard and frequently used scenario in simulations and for good reason. It's often clogged with boats and very large pieces of ice in the late winter and spring. Can software account for all that? That was an extraordinarily lucky save – very light traffic on the river and no floating ice. Also, the plane malfunctioned when it hit the river, a very under-reported story. Both engines are designed to sheer off in a water landing but only one did, that's why it twisted on contact and one side started sinking. The real miracle of that water landing is that the design failure didn't rip the aircraft apart immediately.
The Dish 