Jim Logajan writes:

For example, addition of some machine intelligence in the flight control
systems that takes into account not just pilot demands, but limits to those
demands imposed by the current flight regime, and is active through all
phases of flight so that it aids and/or limits controls to controllable
regimes.

Automation has its own universe of failure modes that is just as difficult to
eliminate as human error (because it is derived largely from human error), and
digital systems in particular have catastrophic failure modes that human-based
systems do not share. So be careful what you wish for.

In any case, Airbus is trying to embrace the philosophy you espouse, not
always successfully.

And to a significant extent, the greater the automation, the less interesting
the activity. It might be pragmatic to automate commercial flight, but flying
for pleasure would probably suffer from excessive automation.

The statistics currently indicate a greater probability of human
failure than machine failure, so this seems likely to yield a net reduction
in the accident rate.

There are no heavily automated systems to compare to, and existing automation
systems have not been analyzed in detail, as far as I know.