The flight log or the FAA etc. doesn't distinguish between AP on and off flying.
Normally you fly manually until flap retraction after take off and the last 1000ft before landing. Of course you can fly the whole approach manually but concentraion goes down rather fast.
Interestingly a few surveys found out that the worst time to switch off the AP/AT is at the outer marker altitude. Either way earlier e.g. at 3000ft or way after that e.g. 1000ft.
1000ft is a good altitude because you have time to adapt to the actual CG, attitude, thrust required etc....
10hrs flight = 5min maximum of manual flying (now you know why I switched back to short/medium range from my longrange career)
Nevertheless even if on AP/AT you have to monitor both very close and check for the least even slighty unusual response from the AP/AT.
Perhaps the only good that can come out of this mishap is to force the airline industry to start taking more seriously the complaints of pilots that there has been too much reliance upon automation. I think this mindset has also crept into airliner design as there has been a trend toward the automation taking more of an active and unoverridable role in flight control. These changes were made due to the analysis of pilot error mishaps in the past, and the engineering conclusion that by automating processes deemed mundane, pilots would be in less a position to allow human error to turn an otherwise properly functioning flight into a deadly mishap.
On paper, it sounds like good logic. But, as with all things, a good idea taken to an extreme often causes problems. If relied upon on mass scale, are computers really better than humans? Or, do mechanical systems ultimately remove the ability of a human to exercise judgment in an extreme situation. I remember Sullerger's heroic ditching in the Hudson. What if automation made it impossible for him to maintain the deck angle he had to enter the water at to prevent catatrophic airframe damage?
But, the third rail of this negative trend is that national flight authoritities, such as the FAA in the United States, have also taken hold on this idea of automation supplanting the role of pilots to fly the aircraft. The classic joke, that certainly isn't a good joke in light of this tragedy, is that the future airliners will have a cockpit built for a single pilot and a dog. The pilot to monitor the systems, and the dog to bite the pilot's hand if he tries to manually fly the aircraft!
You combine this with the human nature to lose sharpness over routine tasks through repetition, and suddenly the very same human behavior problem that so often led to mishaps when perfectly working aircraft were flown into terrain becomes the model where automation that doesn't work as expected -- whether by bad human inputs or malfunction -- goes without notice until too late to prevent tragedy.
It seems to my eye that the solution to this quandry is for the airline industry to assign a series of mandates for manual control of jets during critical phases of flight, with a disciplined requirement of additional pilot oversight to avert mishaps that could result. Further, make these mandates scheduled so that all pilots have to log a certain number of manual controlled events per semi-annual period, and making it a currency item, done at least once every month. Further, airlines should revamp their operating instructions so that pilots are given a tighter series of monitoring protocols that would treat the automation much like an additional human pilot, subjected to the same degree of scutiny that any human pilot would be subjected to when flying the procedures manually.
Ken
