A robot would not have landed in the Hudson unless he was programmed to do that. The MAX disasters are a perfect example of how how well programming airplanes works. We’d probably have had many more MAX crashes if not for human pilots. I still don’t understand what problem replacing proven,safe, and cheap human pilots solves by trying to cut us out, other than self-absorbed paranoid pilots who think the whole world is out to get them!

We are a funny bunch for sure.

Who said anything about human induced? A malfunction by definition isn't pilot induced. That's what makes it a malfunction. And if you really live with this sentiment, you should quit flying and let a more competent and confident human take your place.

The latest trend of crashes proves one thing clearly. Ab-initio programs are detrimental to safety. In other words: less experience means more crashy. That is a case where you can argue the weakest link is the pilot. And even those crashes had malfunctions. How would an aircraft with no pitot data land itself? (Air France). Even in the future? Yet it was totally flyable by an experienced pilot. The one who crashed it was a 1500hr ab-initio baby.

Humans the weakest link? Your copies of William Langeweische articles must be really stuck together.

What happens when there is cabin fire, and the nearest airport is a 8,000' GA field with no instrument approaches. Will hal be able to make that work?

How about when a main landing gear fails to lock. Will hal do a low approach to let the tower check out out? Will it troubleshoot until bingo fuel, ignoring the fact that the bingo was based on the lower fuel burn of having the gear up? Will it pick a longer runway, touching down on the side of the good bogie?

It does not matter.
At some point in the future when the technology will allow HAL to perform with demonstrated capabilities better than a human, there will be pilots that vehemently object to HAL taking over and will do all they can to prevent it.

Sounds like the early resistance to automatic landings, now we simply trust HAL to land 300 RVR all day long. If you were to try to explain the concept of an alert height vs decision altitude 50 years ago (the 737 was already flying, but I digress), you’d get some crazy looks. HAL is also able to fly the wind shear recovery on many planes. At this point we are still needed to monitor and potentially take over, but there will be a day when we’re not. It won’t be in our careers, but the day will come.

If the premise that humans are fallable and the weak link... aren’t humans the one programming Hal? Ergo isn’t Hal susceptible to the same human failings? Only now there isn’t another layer in the cockpit to protect Hal’s mistakes or shortcomings.

Oh by the way, there’s no such thing as a hack proof network... so the idea of remote control goes out the window.

The next argument is for single pilot ops? We went from four engines, to three to two for the same reasons we went from four pilots to two... automation and technological advances, gains in efficiency. Why don’t we have single engine airliners?

We’d still need to pay some minimum wage intern to ride around to wake us up when we doze off. This job gets pretty boring as is, sitting there by yourself would be mind numbing.

Can you imagine a four day solo trip? At least a slam clicker is there to talk to in cruise.

True. Based on the fact we are still flying planes designed 50+ years ago, I’d say we are 100+ years from AI planes. (Not counting short distance drones)

I don't know of many 8,000 ft runways with no approaches. And actually Garmin has already designed an emergency AP function that will recognize an incapacitated pilot and will land the plane at the nearest suitable (it looks at weather and runway performance) airport and will notify emergency services. It requires no ground navigation soruces, so it can conduct the equivalent of a CAT III approach to any runway in its database.

I sure hope not. I am sure the computer will be smart enough to know that low approaches provide no useful data since the tower controllers are not trained in evaluating landing gear failures from a mile away while the aircraft zooms by at 160 mph. And either way, the computer will run the same checklist whether tower says, "The gear looks like its down" or they say, "The gear is partially down"

The answer to all of these is yes. Humans were able to write all of those QRH procedures, so humans can also program those procedures in code. Whatever techniques you were taught can also be programed with the benefit of perfect recall and execution.

Sure we can try to come up with all the possible scenarios where a human pilot is better, but the programers will think of the same scenarios and will program the same response. There WILL be unthinkable scenarios outside the limits of the programming, but then you have to wonder if the average pilot would do better and even if they could, odds are the safety record on a whole would be better than human pilots. 1549 is a perfect example. Sim tests showed that the flight COULD have returned to LGA if you remove human reaction time. Lots of the pilots who attempted this failed however. Even though the crew of 1549 did an amazing job, a computer could have done better simply because it would have almost no reaction time and could troubleshoot the problem while executing a perfect flight path adjusting for the exact gross weight and winds.

But there is no need to worry, pilotless planes won't be a threat because to make a pilotless plane that is better than a human pilot would require lots of money and more importantly, it would require the entire world to rebuild the airspace system. Considering how long it took the FAA to integrate GPS, the ADS-B out debacle, and the fact that some countries still insist on using meters I'm not too worried.