Look at the autopilot issue from a slightly different perspective. What were the expectations of the manufacturer and what now is allowed by the certification authority with respect to autopilot use?
Auto pilot use is mandated for most Cat 2 / 3 operations; for these, a specific design of autopilot is used, these systems do not or must not fail (at least not very often). Some RNP operations require autopilot use to achieve the required navigation accuracy. For all other operations most civil aviation authorities accept that pilots can achieve an acceptable flying standard i.e. safe. It is accept that autopilots will reduce workload, fatigue, etc. But when to use the autos is open to pilot judgement (a skill) and operator policy. The assumption is that either with of without autos the overall operation remains safe.
The semantics or choice of words in previous posts is of concern. When people talk about monitoring an autopilot they should mean monitor the input / output of the system, thus monitor the flight path of the aircraft. Did everyone mean this? It is an old adage that the NFP should monitor the FD during manual flight (older designs of FDs failed often), but the important issue was to monitor what the aircraft is doing – the flight path – how many aircraft hit the hillside with the crosshairs centred? This latter point is of direct relevance to NFPs of HUD drivers. An autopilot does not have a bad day, HUDs should not, but pilots often do. How does the NFP check how the pilot flying the HUD is performing? - Monitor the raw data / aircraft flight path.
Autopilots are designed with differing aims and with varying standards of reliability. Some fail op autos (Cat3b) may only be fail passive in the cruise, thus a different standard of monitoring (workload) will apply. Although the new big jets may have autopilots designed to handle emergency flight situations (engine out), many lesser systems were not. In accident reports it is often these systems that have caught out the crew. What better way of getting the feel of the aircraft by hand flying after an engine failure; pitch trim feel is the feed back for airspeed, similarly, (secondary) so are the roll and yaw forces a cue to what the aircraft is doing. Monitoring is required during a V2 climb, but the urgent actions take place at a predetermined safe point; engine shut down should only require one confirmatory check from PF that the PNF has the correct engine. The subsequent actions are just that, subsequent to the shut down and in a time / priority scale conducive to monitoring safe flight. N.B. Aviate, Navigate, Communicate, Review and monitor (Agitae), in that order. NFP monitors the aviation, navigation, and participates in comms (in other actions take place inter-crew and external).
If a PF or operator believes that a NFP has to spend most of his time monitoring the PF or making MCP selections then there could be a misunderstanding of assumptions made during design and certification of that aircraft. This disparity may imply that the licensing standard of the crew is insufficient for that type of operation, that the individual is ‘just not sharp’, that there is a lack of trust between crewmembers (CRM / human factors), or that the design of SOPs do not match those intended by the manufacturer.
One of the communication gaps in our industry is between what one part of an authority has certificated (manufacture’s design) and what the other part of the authority approves for operation. I agree with the professionals in our industry that the lack of skill (in whatever form) is a threat to the very good safety record that we enjoy; but if we are unhappy about a particular skill then raise the issue within your operation, or with your authority. Unless these concerns are aired formally then any differences in the assumptions made during design and about operating aircraft will go unchallenged, and if so, then there could be a real risk to safety.