There are still studies going forward about letting large aircraft fly with passengers and cargo, with one or no pilots. Some people say it is inevitable, thinking back to the days of professional drivers, or elevator operators, pilots are just an extra expense the airlines can get by without. Technology has improved, it does seem like the pilots don't need to fly the airplane as much as they used to. I've explained how software reliability has improved, and the tools needed to build new autopilots are getting better.
Looking at Air France 447 might be a good place to start. That airplane was apparently flown into the ocean by a pilot, that was confused. The autopilot should have done better, one would think. If the timeline is followed, it will show that the autopilot was confused, and gave up as well (alternate law), relinquishing control to the human pilots. The airspeed sensors (pitot tubes) apparently iced over, causing the automated systems to not have enough correlated information to process the data it had. The pilots left in the cockpit to monitor the systems were not experienced enough to know what to do in this situation.
Talking to most pilots, they will tell you about automation failures all the time. Sometimes stuff just breaks. I know two pilots that were flying back and forth Houston to Austin one day for an airline, and they decided to hand fly the trips and let a flight attendant judge who flew smoother. The next day the one pilot had to take that same airplane from Houston to Orlando, and it turned out the autopilot had failed sometime the day before, and no one knew it broke. He had to hand fly the plane from Houston to Orlando.
The idea of a single pilot in the cockpit is probably just as bad as no pilots. Sometimes pilots have troubles, including health (getting sick, incapacitated, etc), alertness, and just plain forgetfulness. Using a second pilot on the ground might seem like a great idea, one pilot can monitor several flights and "take over" if there is a need. The trouble with the ground pilot is the need for 100% reliable automation, and datalinks. If the part that fails in the autopilot is the same part that the ground pilot will be using for controlling the aircraft (IE servo) it won't help to have someone on the ground wanting to control it, the part is broken for whoever is trying to use it.
Thoughts about using pilotless cargo aircraft are perhaps more palatable, since no one will get killed if the automation fails. That will make sense, if everything being shipped has no value. Things shipped by air a typically worth more than things shipped by truck or train (per Cubic Foot). The pilotless aircraft may still crash, and there will be no heroes on board to steer the aircraft away from the stadium full of people.
Economics probably won't make a pilotless aircraft worth it. Certainly automated systems can be built to make things seem to be cheaper. Certainly pilots are paid well, and have health insurance, pensions and vacation pay that must be paid for by customers. An automated system should eliminate the pilots pay from this picture. There will probably be more maintenance, and a higher price for the initial aircraft purchase. Then the insurance picture may remove all the financial benefits.
Oh, and according to the U.S. Labor Department’s Census of Fatal Occupational Injuries, about 27 people per year are killed in elevator accidents in the US.
What is a life worth?