The problem with fully automated flight is that most criitical incidents develop because of a complex combination of variables which are impossible for programmers to predict, and lead to the need for aircrew to make very fine judgements of a kind which the computers cannot yet make (note the use of 'yet').
An example from the list of FAA anonymous reporting records shows an airliner finding its destination going below minima while on approach, then the alternate closing while in transit, and a further alternate also closing. Finally choosing a diversion field not in the flight plan there were aircraft from all over converging on it. At this field the crosswind made the longer runway marginal, and while on finals the reporting aircraft found that the aircraft ahead slid off the runway, thereby blocking it. The then fuel state necessitated this widebody to land on a short, wet runway at an airport unfamiliar to the flight crew. Getting a computer to cope with that is currently highly problematic.
With the current relationship between people and technology the problem is that of attention. In brief you get more errors on a low tech flight deck, but on a high tech flight deck a higher proportion of the errors made turn into accidents because the humans are slower at recognising them. This is basically because a high tech flight deck is boring to the humans, so their attention is more inclined to wander. Thus when an error occurs it is more likely to be noticed and corrected early on a low tech flight deck than on a high tech one. The paradoxical solution is to incorporate requirements for aircrew attention into high tech procedures.
The other problem with high tech flight decks are the polar opposities of excessive faith in the technology on the one hand, and excessive scepticism on the other. In the former the aircrew can fail to correct a problem because they won't disbelief false readings (of course basic instrument rating training teaches them to do precisely this - believe the instruments and not what your senses are telling you). Excessive scpeticism leads aircrew to lack confidence in the information provided by the technology in a crisis situation.
Both these problems about human perception of the technology means that the training required for flying on a high tech aeroplane involves developing different cognitive modes of problem analysis than is needed for a low tech flight deck. On the other hand low tech aeroplanes keep pilots in touch with the fundamental verities of flight (known traditionally as 'airmanship'), which are precisely what is required in most crisis situations where fine judgments need to be made which are beyond the capability of the technology.
With the present state of knowledge it is not posssible to dispense with either the people or the technology.