Tourist, your earlier view appears to be based on a narrow range of experiences – observations, and the gross assumption that the reduced skill is due to automation.
The retention of high skill level requires practice, but even without practice not all skills will be lost (e.g. riding a bike). The vast majority of pilots, including those you observed appear to have sufficient skills to fly safely – undertake the tasks expected in operational situations.
It would be better to consider why those few pilots flew as they did. Perhaps the problem is not with what you observed, but the process of training; why didn’t the trainer/checker intervene, what did the operator know, the interpretation of regulations, what was the organisation’s attitude, and did the regulator have oversight of this. None of which involves automation or dependency.
The regulators normally consider situations that require us ‘to fly raw data/ old school navaids/ no flight directors/ manual thrust, etc,’; this is often based on probability, which is more suited to the certification process, opposed to human behaviour which entails judged risk assessment. A single observation or accident analysis may not justify re training all pilots, particularly if the required outcome cannot be assured. It might be better to teach pilots how to identify and avoid those situations requiring flight with ‘a limited panel raw data non precision approach in a limiting crosswind when the toys fail’ than expect them to retain a rarely used skill. Thus the safety task is to review our expectations of pilots in today’s operations and not in those which we remember.
The nature of modern operations is that there are few cause-and-effect situations; the range of safety intervention requires careful consideration, without basing them on a miniscule data set. This could be aided with a wider view of human performance and use of other sources for evidence such as the process of training and what happens in normal operations.