There's been a similar argument over aircraft automation in the "Airbus versus Boeing" design philosophy, as I understand it. (Not a pilot so this is my opinion more from the prospective as an occasional software designer.)

The argument goes that Boeing's automation helps the pilot, but it's ultimately the pilot's aircraft to control. Airbus replaces much of the normal flying by computer and can, every now and then, put the pilot into a situation where he has less than full knowledge of how the situation developed or even what the problem is, but where a correct diagnosis and fix must be applied in seconds or else the aircraft will auger into a hole in the ground.

Now in defense of Airbus approach, overall safety records between similar vintage Boeing and Airbus craft is similar, so we are concerned with relatively rare events.

Still, the best studies of man-machine interaction show exactly what Karl has said, absent a 100% automated environment where there is no MMI, the optimum performance and safety point is where the machine helps the operator, and never leaves him in a position with inadequate reaction time or incomplete or misleading information to correct a problem.