Will robots/AI replace human pilots?

Sully's street involved stick and rudder flying that comes from practice, experience and feel that involves, sight, feel, feedback and is different every time. The kinds of things that don't lend themselves to automation regardless of processing power or memory quantity.

Let's just agree to disagree. If I am ever a passenger in a plane in the situation like the Hudson, I will take the old school, experienced, savvy stick and rudder pilot and you can have the automation.

I would too, every time. The problem as I see it is, WE don't get to choose or know who is in the cockpit, the pilots background (and wouldn't THAT be cool if we could find out the PIC's experience and flying hours at ticket purchase time? I know, never happen, but it would not be impossible and would make it quite apparent which airlines put safety first, and they'd compete for the best pilots. Think of that!)

And there is a range of types and people in the employ. I've seen enough of the "Air Crash Investigations" and reading stories, that some are the kind you mention, but some airlines try to hire on the cheap, and some pilots see it as a job, where others are involved in flying GA and keep up their skills.
 
I would too, every time. The problem as I see it is, WE don't get to choose or know who is in the cockpit, the pilots background (and wouldn't THAT be cool if we could find out the PIC's experience and flying hours at ticket purchase time? I know, never happen, but it would not be impossible and would make it quite apparent which airlines put safety first, and they'd compete for the best pilots. Think of that!)

And there is a range of types and people in the employ. I've seen enough of the "Air Crash Investigations" and reading stories, that some are the kind you mention, but some airlines try to hire on the cheap, and some pilots see it as a job, where others are involved in flying GA and keep up their skills.
Not sure what you mean by they see it as a job while some are involved in GA?
 
For every Qantas incident that COULD have killed people, how many incidents DID kill people as a result of human error?
You who are on the technology conquers all side of the equation keep repeating this as gospel, when it's not the whole equation. The other half of that equation is "How many times have humans kept automation from killing people?" This is the question that you all are not addressing. We on this board who fly passengers and freight around on a daily basis, using the most modern technology in airliners today are telling you that there are many, many times that human intervention saves a flight from becoming a news story. The problem is that that never becomes a news story. All it becomes is two sentences in a aircraft maintenance logbook and a story at the bar. Everyone sees when the human element fails. It's the front page of every paper and the lead story on the 11:00 news. Very rarely (like the Quantas story) do we hear about when technology goes awry and thankfully, there's a human in the loop to correct the issue.

Maybe you guys skipped over it; maybe you're ignoring it, but a few pages back I wrote a post where I detailed just three issues in Boeing 777's dealing with automation/programming errors that if a human wasn't there could have been catastrophic. Boeing is dealing with these issues right now, but haven't been able to solve them. Their interim solution in all cases is to have the human intervene until they can figure out and solve the problem. If the human wasn't there, what would the solution be? Ground the entire worldwide fleet of 777s until they can debug and reprogram all aircraft?

There's your answer as far as automation it concerned. If a computer kills people .00001% of the time and a human kills people .00002% of the time, the computer is safer. Period. Full stop. End. Fin.
Do you really think so? Are you one of those "just think of the children" types? At what cost are you willing to save .00001%? Trillion dollars? Ten trillion? How about if we don't allow checked baggage and make everyone fly naked. No more terrorist threat! Just think of the people we would save. Let's get rid of cars! Make everyone ride bicycles! If bicycles just saved .00001% of the population, it'd be worth it, right?

It's a ridiculous notion that we'll be able to save eveyone, in addition to the fact that I don't necessarily agree with your initial premise that computers will kill less people than human error would. That's not proven, and I think quite the opposite (see my first paragraph).
 
continue with your write-ups....the ops guys do spread that info to folks who do the designing....trust me on that. :D

Sooner or later....those bugs will be solved....and the human will once again become bored with his job.
 
continue with your write-ups....the ops guys do spread that info to folks who do the designing....trust me on that. :D

Sooner or later....those bugs will be solved....and the human will once again become bored with his job.
Why were the bugs there to begin with? That's what I want to know. Are programmers/engineers just slacking now, writing bad code saying "F it... there's a dude sitting there." When they're designing an autonomous airplane are they finally sit up straight and say "let's get to work now boys, this time it counts!"?

Why wasn't this stuff cought before it was installed in a production airplane carrying passengers around?

Autothrottles that decide to go to takeoff thrust on their own on the ground? That's a great feature! Why was that put in the code? Aircraft dumping all FMS data when updating the descent winds? Super... good luck making that crossing restriction.

I want to know why I have to deal with this stuff in a modern airliner. Why were these errors made and why do I always have to save the public from all these engineering and programming mistakes!?
 
continue with your write-ups....the ops guys do spread that info to folks who do the designing....trust me on that. :D

Sooner or later....those bugs will be solved....and the human will once again become bored with his job.
The model airliner that I fly first entered service fifty years ago. How much longer do the engineers need to squash those bugs?
 
.....I want to know why I have to deal with this stuff in a modern airliner. Why were these errors made and why do I always have to save the public from all these engineering and programming mistakes!?
because you are our super hero human that will fix things....until the technology is fixed. :D

So....is the 787 doing all that stuff....or the 737 Max?
 
because you are our super hero human that will fix things....until the technology is fixed. :D
Im not a super human. I'm not even a super pilot. I have limitations that I understand. I just wish some of you understand that there are limitations to technology as well. I'd like to think that I have a symbiotic relationship with technology at work. I'm not some Luddite that eschews technology. I actually embrace it, work with it. I understand it's limitations and I use it to reduce my own.

So....is the 787 doing all that stuff....or the 737 Max?
I don't know. I don't fly those aircraft. I flew (up until a few months ago) brand new B777s that were rolling off the assembly line. And I will tell you that if the RNAV in the 787 and 737 Max is the same as the 777 and 757, pilots in those aircraft are complaining just as much about airplanes that can't make speed/altitude crossing restrictions without their intervention 50% of the time.

Inn full disclosure, McDonnell Douglas's version of VNAV (PROF) in the MD-11 was usually spot-on and 1000 times better than Boeing's VNAV.
 
You who are on the technology conquers all side of the equation keep repeating this as gospel, when it's not the whole equation. The other half of that equation is "How many times have humans kept automation from killing people?" This is the question that you all are not addressing. We on this board who fly passengers and freight around on a daily basis, using the most modern technology in airliners today are telling you that there are many, many times that human intervention saves a flight from becoming a news story. The problem is that that never becomes a news story. All it becomes is two sentences in a aircraft maintenance logbook and a story at the bar. Everyone sees when the human element fails. It's the front page of every paper and the lead story on the 11:00 news. Very rarely (like the Quantas story) do we hear about when technology goes awry and thankfully, there's a human in the loop to correct the issue.

Maybe you guys skipped over it; maybe you're ignoring it, but a few pages back I wrote a post where I detailed just three issues in Boeing 777's dealing with automation/programming errors that if a human wasn't there could have been catastrophic. Boeing is dealing with these issues right now, but haven't been able to solve them. Their interim solution in all cases is to have the human intervene until they can figure out and solve the problem. If the human wasn't there, what would the solution be? Ground the entire worldwide fleet of 777s until they can debug and reprogram all aircraft?

Do you really think so? Are you one of those "just think of the children" types? At what cost are you willing to save .00001%? Trillion dollars? Ten trillion? How about if we don't allow checked baggage and make everyone fly naked. No more terrorist threat! Just think of the people we would save. Let's get rid of cars! Make everyone ride bicycles! If bicycles just saved .00001% of the population, it'd be worth it, right?

It's a ridiculous notion that we'll be able to save eveyone, in addition to the fact that I don't necessarily agree with your initial premise that computers will kill less people than human error would. That's not proven, and I think quite the opposite (see my first paragraph).
I think some folks are not paying enough attention to the experiences of the airline pilots such as yourself who have spoken up. As a retired engineer, I have learned to respect the views of people who have used the results of my work in their profession day in and day out.

You are correct that the comparative safety records are not the only criterion in deciding when a new technology is ready for the marketplace. As you point out, economic feasibility and end-user acceptance must also be taken into account.

I don't say that fully automated airliners will never pass all three tests. My position is that we just don't know. I do think that many are underestimating the difficulties.
 
Last edited:
Why were the bugs there to begin with? That's what I want to know. Are programmers/engineers just slacking now, writing bad code saying "F it... there's a dude sitting there." When they're designing an autonomous airplane are they finally sit up straight and say "let's get to work now boys, this time it counts!"?

Why wasn't this stuff cought before it was installed in a production airplane carrying passengers around?

Autothrottles that decide to go to takeoff thrust on their own on the ground? That's a great feature! Why was that put in the code? Aircraft dumping all FMS data when updating the descent winds? Super... good luck making that crossing restriction.

I want to know why I have to deal with this stuff in a modern airliner. Why were these errors made and why do I always have to save the public from all these engineering and programming mistakes!?
The most likely answer I've seen to those questions is this post from MAKG1:

https://www.pilotsofamerica.com/com...eplace-human-pilots.90758/page-5#post-2282927

Here is an article on the "NP-complete" issue that he referred to:

https://en.wikipedia.org/wiki/NP-completeness

This portion of the introduction seems applicable (boldface added by me):

"Although any given solution to an NP-complete problem can be verified quickly (in polynomial time), there is no known efficient way to locate a solution in the first place; indeed, the most notable characteristic of NP-complete problems is that no fast solution to them is known. That is, the time required to solve the problem using any currently known algorithm increases very quickly as the size of the problem grows. As a consequence, determining whether it is possible to solve these problems quickly, called the P versus NP problem, is one of the principal unsolved problems in computer science today."​
 
keep writing up those problems....fill up the log, and eventually it will be addressed.
 
So....is the 787 doing all that stuff....or the 737 Max?
The 737 MAX is just lipstick from an automation perspective. Bigger LCD screens. The systems are substantially unchanged since the original model fifty years ago. Still no EICAS, just the antiquated RECALL panel which was designed to eliminate the flight engineer. Still lacks the systems automation that has been in the bigger Boeings since the 757/767 in the 1980s.

The big improvement is the more efficient engines.
 
Who will write up the problems once the human pilots are eliminated? :devil:
The bigger issue is that he is expecting all problems to be known ahead of time. To fully automate an airliner the automation system must solve problems that have not been anticipated.
 
The bigger issue is that he is expecting all problems to be known ahead of time. To fully automate an airliner the automation system must solve problems that have not been anticipated.
Na....the human will discover it....just like now. But if no one sez anything who's gonna do anything about it?
 
Not sure what you mean by they see it as a job while some are involved in GA?

Sorry, I didn't write that very well. I mean that I get the idea that, maybe outside the US more than it it, but that there are pilots that decide to learn to fly commercial not so much for the love of flying as for the money and career. Other pilots love flying, and when not flying for their job, fly privately on their off time, etc.
like any job, there are people that maybe are not keeping up their skills, and others that are just plain pilots through and through?
 
Done in perfect weather over flat terrain, of course.
 
The most likely answer I've seen to those questions is this post from MAKG1:

https://www.pilotsofamerica.com/com...eplace-human-pilots.90758/page-5#post-2282927

Here is an article on the "NP-complete" issue that he referred to:

https://en.wikipedia.org/wiki/NP-completeness

This portion of the introduction seems applicable (boldface added by me):

"Although any given solution to an NP-complete problem can be verified quickly (in polynomial time), there is no known efficient way to locate a solution in the first place; indeed, the most notable characteristic of NP-complete problems is that no fast solution to them is known. That is, the time required to solve the problem using any currently known algorithm increases very quickly as the size of the problem grows. As a consequence, determining whether it is possible to solve these problems quickly, called the P versus NP problem, is one of the principal unsolved problems in computer science today."​
Yes. This is just a formal way of saying (with evidence) that more computing horsepower will not solve every problem. There are many problems that are complex enough that it is not practical to solve them completely in any reasonable amount of time. Encryption key recovery/decryption and system testing(of complex software/hardware systems) are examples.

Until you test and prove "correctness" for an entire integrated system(which is not practical for modern complex systems, see above), that system will contain bugs/errors that will arise under certain operating conditions. I have worked in software engineering and industrial automation for most of my career. I have seen many unexpected failures, some of which we unable to explain or repeat... :). I am sure that the capabilities of AI systems will continue to increase exponentially and that autonomous aircraft will be developed. The question is: at what point do we decide that a human operator/pilot to oversee system operations (and can intervene/save the day in case of systems failure) is no longer a useful, safety-enhancing, part of the system. I think, for passenger carrying aircraft, a human pilot will be needed for a long time yet.
 
Virginia's gov the first official to fly in an automated aircraft.....

Between this video and the 737 one, I just feel better about autonomous aircraft taking my job in the foreseeable future.

The headline for this should read "Virginia's gov flies in an airplane on autopilot." Aircraft have been automated for a long time. Let me know when the headline reads "Virginia's gov the first official to fly in an AUTONOMOUS aircraft."

And the 737 video is laughable. The headline reads "Robot lands 737 By Hand." It did nothing of the sort. The robot twisted a couple of knobs, moved the gear and flap lever, and actuated the thrust reversers. Hardly landing "by hand." And i was watching it, all I was thinking was, holy cow this thing is taking forever to get this stuff done, then I saw in the upper right part of the screen "4x Speed." Wow. I know this is proof of concept or something like that, but it's got a long way to go to rival even the worst First Officer I've seen (me). I wonder how all that knob twisting is going to work in moderate to severe turbulence. And the other question I had was if it was deciding when to slow, deploy flaps, gear, or was it being fed commands?
 
I'm okay with automation in the cockpit. Personally, I will never be a passenger in a plane that does not have a pilot, nor a car that has no driver. Automation is fine as a method of increasing safety and efficiency of operations, but no device designed, built, and programmed by human beings can ever be 100% reliable.

In the future, even if the only "flying" the commercial pilot ever does is in the simulator, I still want him up there in case the gizmos fail. Otherwise, they ain't gettin' my money.
 
I will never be a passenger in a plane that does not have a pilot, nor a car that has no driver.

Sure you will. Ever use your cruise control and follow the directions coming out of your cars GPS? You are half way there already.
 
I'm okay with automation in the cockpit. Personally, I will never be a passenger in a plane that does not have a pilot, nor a car that has no driver. Automation is fine as a method of increasing safety and efficiency of operations, but no device designed, built, and programmed by human beings can ever be 100% reliable.

In the future, even if the only "flying" the commercial pilot ever does is in the simulator, I still want him up there in case the gizmos fail. Otherwise, they ain't gettin' my money.
+1
 
I'm okay with automation in the cockpit. Personally, I will never be a passenger in a plane that does not have a pilot, nor a car that has no driver. Automation is fine as a method of increasing safety and efficiency of operations, but no device designed, built, and programmed by human beings can ever be 100% reliable.

In the future, even if the only "flying" the commercial pilot ever does is in the simulator, I still want him up there in case the gizmos fail. Otherwise, they ain't gettin' my money.

^^^Same

Sure you will. Ever use your cruise control and follow the directions coming out of your cars GPS? You are half way there already.

lol no
 
And if you jump into a Tesla, Apple or Google car or an Uber or Lyft autonomous you are already 100% of the way there. City buses are being converted to autonomous electric too. How long before you hop into an autonomous air taxi to get to and from an airport and to a hotel? Five years maybe. How long before the commerical flight you are taking has a right seat only human babysitting the automation? Twenty years max.
 
And if you jump into a Tesla, Apple or Google car or an Uber or Lyft autonomous you are already 100% of the way there. City buses are being converted to autonomous electric too. How long before you hop into an autonomous air taxi to get to and from an airport and to a hotel? Five years maybe. How long before the commerical flight you are taking has a right seat only human babysitting the automation? Twenty years max.

LOL. Ain't nobody going to wear three stripes to watch a robot work. "Right seat only human"... you'd be hilarious if you weren't so clueless. You may be right about single pilot cockpits someday, but no organization being as PC as they all are these days would tell the human they are "first officer" to a robot. ROFLMAO. Human egos won't allow it.

We're living in the age of companies telling call center drones that they're "family" and pretending they're all one big "team" and having rah-rah meetings where they get workers to clap and cheer at little propaganda meetings like a "family" would normally lay off someone and put them in the street whenever the numbers don't work for the patriarch/matriarch, and they throw little ice cream socials with games like it's a freaking family picnic.

Tell the human they're wearing three stripes... LMAO... right.
 
I get a little worried about people growing faith in computer systems in areas which are not controlled such as on the roads and in the sky. On the roadways mainly because of weather, conditions, and especially the fact that it is often a crowded environment, with all kinds of other variables such as pedestrians, animals, etc.
In the skies because of weather conditions, and the fact that a computer can only "know" what it can read from sensors for the outside world.

Faith in programming and testing is not warranted. Read up on systematic failures in our history, or the problems of testing. If a computer has bad data it is garbage in garbage out. A blocked pitot, a miscalibrated sensor, a sensor that fails, etc make a computer make crazy decisions. Imagine you made a simple calculator, not even scientific with cos, square roots, etc. and wanted to test it. Even if you programmed a robot that could hit the keys at a rate a hundred or more times a human, to test every combination of numbers adding, subtracting, multiplying, dividing, it would take a century to test them all. They test representative and assume all the others are correct. On top of that, who checks the checker? There may be glitches in certain combinations.

It's humans that program, and decide what to test, and how, and then how to verify the results.

Computers have no "will". They ar programmed to tasks, but the pilot has skin in the game. Humans have much more adaptability though we make mistakes, than programs.

Then there is connectivity. Security, vulnerability to hacking and outside forces wanting possibly to crash a plane.

There are many stories I've seen on "Air Crash Investigaions" where pilots, with damages to the airplane (through explosions, strikes, failures of components) that managed to deal with, fly, and land planes that seem almost impossible and I am one hundred percent certain no program would have been able to tackle. It's happened.

I'm no Luddite, but sometimes people out way too much faith in technology. I'm a programmer, and I have most definitely seen a huge degradation in engineering and programming examples now that I thought not even possible. New updates in my tv box, it's like a TiVo and receiver, etc. are like a textbook of bad programming and user interface.
My wife bought me a DAB alarm clock recently. How badly can one design an alarm clock? To depths I hadn't thought possible. LCD display straight on looks fine, but the light coming off any angle, lights up the room like stadium lights. The set alarm function is hidden below about four submenues, which to access you have to hit combinations of keys. You can see the alarm is set, a cute little alarm bell is shown, but to see what it is set to, you hit two of three buttons at the same time, and then must scroll (another button) to alarm and choose it, but must do this within like two seconds or it just goes back to the main time screen.
On the light thing, I did find three settings for intensity. One was brightest, next was really about the same with sidelight, the third and lowest...off. So no display at all.
This was a hitachi.

I think we are going towards ever more sophisticated logarithms, and concepts, but that are beyond our ability to test, and there is a steep decline of "common sense" and best practices, and design going on now.

When it comes to automobiles too...they've had very bad experience which should have been obvious if thy had thought at all, with "autopilot" on autos where the driver is still "responsible" ultimately. That is insane. Drivers that believed the car was fully autopilot, have crashed. They then blame the driver. The thing is, their model was that the autopilot would drive the car, but the driver had to keep aware and be ready to take over if needed. Obviously you are more in the game if you are driving. In fact their model seems to be the driver had to be more aware all the time than if they were driving. And of course there would be a lag as the driver (assuming he isn't asleep, reading a newspaper, skyping or something) waited to see if they should take over.

Ultimately, driving is a thing you need to keep doing to keep up proficiency, and also, a pure pleasure if you are a good driver to smoothly brake, keep speed constant, drive well. Flying is even more so. Pilots dig flying and the best ones aim for perfection right? Make it mundane, and something anyone can do, and you aren't going to keep the good pilots, and then where are we when systems fail?
 
Since this thread seems to be about predicting the future, I predict that LongRoadBob's very valid concerns above will be glibly dismissed.
 
Back
Top