I think that would set a dangerous precedent. Systems mess up in all kinds of strange ways. The human pilot needs the final say, especially in an emergency situation.

I think they were saying the autopilot only kick in if the pilots fail to acknowledge it.

No. Computers aren't magic in that way, and we leave those types of decisions up to pilot's for a reason.

With any advanced system, we use automation to decrease the complexity of a task by which to act as a throughput modifier by decreasing barrier to entry for safe pperatiom.

If you start throwing in computers which fight the operator at every step, you'll start seeing people (1) move away from the system, (2) subvert the system, (3) start trying to make a buck off of selling systems they claim will fix everything, while not really dealing with the issue.

Plus, computer vision just isn't there yet.you're talking about giving a machine the ability to integrate a whole lotus different streams of data and reliably making a decision to veto the pilot's instructions.

Air plane auto-pilots keep the plane on a specified course; they don't do obstacle avoidance. Planes aren't even programmed to avoid mountains, let alone buildings.

So, you'll put locations of all things that plane is not supposed to fly into on the plane? Also you will program it to automatically fly around? #whatcouldpossiblygowrong?