etudiant, a pedantic clarification of your design concept.
The design requirements for large aircraft specify limits for autopilot performance (FAR / CS 25).
Generally any single autopilot malfunction (not the same as A/T) should not result in the aircraft being ‘significantly’ out of ‘trim’ - vs ‘huge’ and ‘unstable’.
The descriptive term ‘significantly’ involves subjective interpretation of what a suitably qualified pilot without need of special skills / knowledge or strength will be able to manage.
The aircraft requirements do not specifically consider combinations of systems, except within the overall requirement for an adequate level of safety (usually a numerical probability).
The level of safety for interacting malfunctions - A/T or engine, and autopilot, often depend human intervention (subjective probability).
With A/T malfunction the autopilot may struggle. However, because the resultant flight path is not caused by an autopilot malfunction, the certification aspects of the combined effects with engine malfunction may use pilot awareness and action to meet the required level of safety.
Whereas aircraft have to meet design requirements (hard science), no such requirement exists for humans.
The human contribution is usually judged subjectively (soft science) - non numerical probability, a certification opinion, an assumption about human performance in a given situation.
Many accidents and events involve the human contribution. Increasingly outcomes are inappropriately attributed to error or individual’s ability; neither has meaning in certification, nor in operational investigation.
All that might be concluded is that there appears to be a mismatch of certification expectations of crews’ contribution, and that achieved in the situation encountered; where the situation must also consider the technical malfunction.
We cannot tell if certification has misjudged crews’ abilities to respond, or that crews’ abilities were sufficient, have changed, or situations have evolved, more prevalent, including a range of interacting contributing factors, aircraft / system age, maintenance, airworthiness safety monitoring.
The interaction of quantitative and qualitative assessments cannot be judged in isolation because of the complexity of the overall situation: -
‘ seriously askew’ - yes, a mismatch between the situation as perceived by the crew and subsequent (in)action, and the assumptions of these by certification before the event, or as reviewed with hindsight after the event.