I wonder if a human is in the loop. Obviously the software is hardly ever used (a good thing), so you wouldn't need many humans available. If communication is possible, wouldn't you hand control to a pilot on the ground?
I don't know that they could actually fly the plane - is latency too high for landing? - but they could make all the decisions and communicate with air traffic control, other planes, and the passengers.
The problem is every aircraft model flies differently. The remote pilot would need to be familiar with that particular type of aircraft to safely land it.
I'm thinking of higher-level contributions such looking at the weather and saying 'fly to this airport and use this runway'; or asking the passenger, 'what does this gauge say?' or 'look at the left engine; what do you see?'; or talking to air traffic control.
One major difference is if a uav crashes no one dies. But in china there is apparently now a commercial pilotless flying ev taxi service - which is autonomous with a human on the ground in the loop as you are suggesting.
Remote piloting for landing an aircraft that size is problematic because you need more sensors on the aircraft plus a reliable, high-bandwidth, low-latency data link. That doesn't really exist in most places. When the military lands something like an MQ-9 Reaper they typically hand off control to a pilot located within line-of-sight right at the airfield. That obviously isn't practical for civilian general aviation.
I don't know that they could actually fly the plane - is latency too high for landing? - but they could make all the decisions and communicate with air traffic control, other planes, and the passengers.