The fact is that I do not understand the mindset; if he had to reset the autopilot three times, why did he then engage the GA mode (a four axis mode). Having engaged the GA, why did he not trust it until it was established that it was not doing what it was expected to do. All this given that there were issues compounding the situation - including an aircraft (which is not the best performer in any case) that weighed more than the documentation indicated.
My guess would be that this serious incident once again shows that the level of understand of a pilot for what is a very complex piece of equipment (and algorithm) is indicative of general knowledge among the pilot work-force. In my view that is not solely the fault of the pilots but of a system of training and operations which has hardly changed since the introduction of these very complex systems in the mid 2000s.
Jim