NHTSA is starting to walk back Tesla and its "autopilot" and it's starting to look like a failed experiment. Ironic that he used the word autopilot to sell the idea that it does the driving itself. Last I checked, with THREE autopilots, two highly trained pilots monitoring, stringent monthly maintenance requirements for on-board equipment, and a highly controlled operating environment being managed by ATC, pilot intervention is regularly required. So hey, clearly $50K should get me a magic car that can just handle things.

I'm convinced that the industry hasn't the slightest idea how many seemingly tiny things an on-site human pilot does both in terms of safety and in facilitating operations. Literally hundreds or thousands of tiny decisions are made on a flight (starting well before departure). These decisions are made quickly, accurately, and in a timely fashion because the decisionmaker is in a physical position to gather and process information.

Justification for pilotless flying in the military are completely different. These include a much higher acceptable loss rate, high-g maneuvering, extended loiter time, etc. Airliners (even cargo) are full of live things as well as packages that will continue to require a controlled environment - which essentially eliminates many of the efficiencies of a pilotless aircraft.

I'm sure it's possible to make pilotless flight safe-ish. I don't think it will save anyone money - I think it will ultimately be more expensive and less reliable.