CAN YOU IMAGINE how dependent we have become on computers to do things, and THUS when a NUTTY value comes along the COMPUTER BELIEVES IT AND THEN GIVES UP>
As if the "computer giving up" isn't an implicit assumption from the design team that the pilot will know better to do at that point.
The crux of the matter is that training and improvement continues between human and machine interface so that incidents where the autopilot cuts out aren't "oh
" moment for contemporary pilots.
As a long time lurker, in my observation, post AF 447, PPRUNE accident threads unfold in rather predictable patterns, usually a dichotomy of "airmanship vets" verse "software guys and magenta line pilots (and management)" with each side talking past the other.
I hope that this incident, if found to be caused by poor human machine interface, leads to increased training and awareness of what exactly is happening at the moment of criticality to throw otherwise professional, trained pilots off their game so much to lead to an upset.