Sorry, but this is a very common misconception. Just like many people not familiar with programming think that anyone can make a great software product just by pointing and clicking in some GUI tool. The tools certainly help a lot and makes the job easier, but pilot-less passenger airplanes are still many years away.

And, knowing just how much can go wrong both with software and with flying, I know I'm not going to fly with such an airplane. Auto-pilot and auto-land systems today work well because they do a very narrowly defined thing with great precision, with the amount of software needed kept to a minimum. A software system that would be able to account for all that can go wrong when flying would have to be very complex and difficult to get defect free (and don't get me started on the idea of remote controlling passenger planes).

Right. Except for those times when you need people like Captain Sully to deal with the crisis at hand.

Being a programmer, I wouldn't want to trust my life to a computer program with no human supervision.

You don't pay pilots for the 9999 flights where nothing happens and the autopilot works fine. You pay them for the 1 flight where the autopilot either fails or encounters a situation it doesn't know how to handle, but amortize it over all the ones where it goes fine.

Computers don't right the rules, people do and guess who works in the FAA? Former pilots.

Autonomous drones can already fly but they block off dozens of sq.miles for them to fly. Anyway there is legislation going up to allow drones in controlled airspace. Guess what? It requires each one to get the same type rating and hours as a commercial pilot. Which increases the cost significantly.

Landing is typically done on autopilot these days; it is done manually to maintain a pilot's rating or for exceptional reasons.

Actually, the exact opposite is true - pilots must use autopilot on a certain number of occasions to retain their certification to use it on low visibility approaches, but otherwise generally do so only when visibility is severely restricted.

...and under fair weather conditions. Under more adverse conditions will be done manually.

It's fairly easy to tell a computer, "See that line (runway slope), follow it".

I think the autopilot follows a ILS beacon in. Isn't that why less flights are diverted in poor weather now? http://en.m.wikipedia.org/wiki/Instrument_landing_system http://en.m.wikipedia.org/wiki?search=Autopilot

Landing is almost never done by autopilot. Approach is. When the runway is in sight Pilot Flying has to take over from the autopilot, and gently put the wheels on the ground. Computers are still not very good at it, especially not when there is slightest deviation from the norm (i.e. crosswind).

Autoland is very rare and is either used in ideal conditions or in terrible conditions (but then both aircraft and airport AND flight crew have to be rated to Cat IIIc standard - extremely rare occurrence).

Pilotless military planes are designed to kill people. I would prefer not to use that as a criteria for creating pilotless passenger-carrying aircraft.

Like the Sao Paulo-Paris A330 that crashed in the ocean mid flight? Was on autopilot, wasn't it ?

Thunderstorms and lightning don't care who's driving the plane.