This. The devil is in the details. We have not been on board to hand-fly the plane for a long time. We are onboard to make decisions and provide redundancy.

Can it be done today? Yes

Can it be done today with Equivalent Safety? No. The last 0.001% is the real hurdle, and that's where humans come in. We're flexible and creative, and AI isn't there yet.

Can it be done economically today? No. Too much required redundancy, and too many CANX flights due to WX, MX, computer flashes a code, etc. Frankly cheaper to just pay pilots to do it for the time being.

It will have to be a special-built airliner. Just because Boeing is studying it doesn't mean they're building it, or could build it.

Once you solve all of that, you have to get regulatory approval, and re-design the ATC system. That's 30 years and about a trillion dollars right there.

I understand this is scary if you're an liberal arts major, but anybody with a background in systems engineering, computer science, or even government knows this is a lot harder of a problem than it seems.

People are OK with fatal highway accidents. But they have very low tolerance for fatal airliner accidents (it's a control thing). They will have zero tolerance for fatal accidents involving unmanned airliners. The people who would build, approve, and operate such things know this. They will most likely not launch any half-assed experiments. Long ways to go.

Big ROI on automated trucks (millions of truck drivers).

Big ROI on self-driving cars (billions of drivers, who could make better use of their time watching jerry springer).

Not much ROI on eliminating airline pilots, orders of magnitude less (fewer than 100K airline pilots in the US). But the cost of replacing us is orders of magnitude higher than for automating cars & trucks.