United 767-300 Has Fuselage Buckled by Hard Landing at IAH July 29, 2023

[Astro14]

It’s not commonly known but a heart surgeon will schedule multiple surgeries for a day. His assistant does all the opening and getting things ready. The heart surgeon comes in and does the important part then leaves to go to the next patient. The assistant closes up.

As for emergencies, they do happen. I’ve seen several emergency open heart surgeries performed right in the patients room usually because a graft pops loose. They open them up right there, stabilize things and then head to surgery to finish it up.

In major hospitals one cardio thoracic operating room is left open for emergencies. You might have three heart surgeries going and suddenly a gunshot wound to the heart comes in. A lot of juggling behind the scenes happens.

Those are sequential surgeries. Not simultaneous. You can reorder them in triage, but the surgeon doesn’t have two open hearts at once, unless another doctor is there.

Point is, during simultaneous operations, when something goes sideways in one venue, the others will get ignored.

To their detriment.

Remote pilots. Simultaneous pilots. Are both foolish, risky suggestions.

The only people who cannot see how foolish they are, are the people who simply do not understand what a pilot does.

When an engine fails on take off, for example, you have less than one second to put in the correct amount of rudder before losing control of the airplane, crashing, and losing everyone on board. Remote pilots can’t do that (and the predator is flown locally for takeoff and landing for that very reason) and pilots managing multiple airplanes can’t do that, any more than a surgeon can hold five scalpels and use them all at once.

 

[atikovi]

When an engine fails on take off, for example, you have less than one second to put in the correct amount of rudder before losing control of the airplane, crashing, and losing everyone on board. Remote pilots can’t do that (and the predator is flown locally for takeoff and landing for that very reason) and pilots managing multiple airplanes can’t do that, any more than a surgeon can hold five scalpels and use them all at once.

AI will have learned all this in 5 or 10 years. The flight control system will take corrective action BEFORE losing control. The computer can make millions of calculation a second and prevent anything that would cause the plane to crash.

 

[MolaKule]

AI will have learned all this in 5 or 10 years. The flight control system will take corrective action BEFORE losing control. The computer can make millions of calculation a second and prevent anything that would cause the plane to crash.

Again, we return to the problem of programming the AI, done by humans. I do not think AI will be able to "feel" the aircraft as a human can do, no matter how much feedback is given to AI in an attempt to learn it.

AI cannot replace the "lifetime experiences" of a pilot.

 

[atikovi]

AI cannot replace the "lifetime experiences" of a pilot.

Isn't that the whole premise of AI. That it learns from billions of data points to become smarter and more capable than any human could ever be?

 

[Drew7a]

https://www.space.com/25275-x37b-space-plane.html

I think the next step will be down to one pilot + AI first before ultimately making the leap to AI.

Just like they said in the past you could not do long over water flights with less than 4 engines, then 3, then now we have ETOPS.

Yes human feel is important but like any professional there are excellent Pilots and mediocre ones.

 

[Astro14]

AI will have learned all this in 5 or 10 years. The flight control system will take corrective action BEFORE losing control. The computer can make millions of calculation a second and prevent anything that would cause the plane to crash.

You know nothing about flying, airplanes, or the airline operating environment, yet you make this prediction?

Based on what? Ignorance? A desire to troll? How well self driving cars are working out?

If AI can’t handle cars, it’s decades away from handling airplanes.

Sure, people are working on it, but billions of calculations without the sensors, or inputs, means billions of irrelevant calculations that are no closer to managing the situation than we are now.

AI might assist airline pilots, just as AI might assist doctors, but it’s not replacing them in passenger airliners any time soon, and certainly not in your timeline.

 

[64bawagon]

No insider information on this.

But with that wrinkling- it’s going to be an expensive fix at best and likely written off.

This was a trained pilot at the controls.

Can’t help but think about the thread where most folks without training think they can land an airliner. Sure they can…it ain’t that easy…

Landing is easy, its how youre able to walk away from it that makes all the difference.

 

[Astro14]

Landing is easy, its how youre able to walk away from it that makes all the difference.

Again, no, it isn’t easy, and this proves it.

 

[atikovi]

You know nothing about flying, airplanes, or the airline operating environment, yet you make this prediction?

Based on what? Ignorance? A desire to troll? How well self driving cars are working out?

If AI can’t handle cars, it’s decades away from handling airplanes.

Sure, people are working on it, but billions of calculations without the sensors, or inputs, means billions of irrelevant calculations that are no closer to managing the situation than we are now.

AI might assist airline pilots, just as AI might assist doctors, but it’s not replacing them in passenger airliners any time soon, and certainly not in your timeline.

I said in 5-10 years. Nobody can tell how much tech and AI will have advanced by then. It wasn't long ago that a hard drive was 10MB. Today one high res photo would fill it up. As more powerful computers are introduced, especially quantum computers, the stuff that seems impossible today will become commonplace then.

 

[Astro14]

I said in 5-10 years. Nobody can tell how much tech and AI will have advanced by then. It wasn't long ago that a hard drive was 10MB. Today one high res photo would fill it up. As more powerful computers are introduced, especially quantum computers, the stuff that seems impossible today will become commonplace then.

Mark your calendar. I will retire in 5 years.

Let’s see how well this prediction* ages.

*Based on zero understanding of how airplanes work, the operating environment, and what it takes to fly an airplane. It takes far more than just computation, there is the entire sensing problem that I mentioned previously. Self-driving cars still can’t “see” pedestrians, or the road, reliably.

So, how would the AI driven airplane “see” to to land, to taxi, to avoid other airplanes in flight (as I have done, using purely visual means, when an airplane blunders onto final ). The USN has drones that use automated flight logic, but they don’t have the safety margin for traffic avoidance, nor do they have the ability to taxi autonomously.

Without the ability to “see” AI has no chance of safely operating an airliner, any more than an AI driven robot can do surgery when it can’t “see” the patient, and react to variations in physiology, and changes in the situation. Computation alone isn’t enough.

 

[atikovi]

So, how would the AI driven airplane “see” to to land, to taxi, to avoid other airplanes in flight (as I have done, using purely visual means, when an airplane blunders onto final ).

Radar, lidar, IR, Doppler, plus some yet undeveloped technology.

 

[GON]

One only need to study United flight 232 to know drone controllers will not replace airline pilots on commercial aircraft.

Yes AI will make many of our life's miserable when trying to get customer service, and we will be subject to fraud from ai. But ai can't do one thing humans can do, regardless of what the California deep pockets tell us. Ai can't critically think.

https://en.m.wikipedia.org/wiki/United_Airlines_Flight_232

 

[wwillson]

Who programs the autopilot?

Who configures the airplane?

Who selects the runway, flap setting, approach speed, autobrakes?

How does the airplane taxi once landed?

All of those functions are currently done by a trained crew. Autopilots cannot do that. Remote pilots cannot do that.

You need to understand how an airplane actually operates before making suggestions on how they can be flown.

@Astro14 is spot on

Also consider:

Who lands the airplane when the autoland system says, "Not going to land in these conditions, sorry". Yes, we had the autopilot disengage on short final, because it couldn't handle the 40kt crosswinds and turbulence. A human had to land, because the machine couldn't. We did landings over and over with the same conditions. These were very difficult landings and with so much crab at touchdown, the landings aren't pretty, but the humans got it done when the computer couldn't.

Without a human in the cockpit, who uses critical judgement when things go wrong?

 

[Astro14]

Radar, lidar, IR, Doppler, plus some yet undeveloped technology.

Oh, I see… “Doppler” is a physics property, not a technology, so, “Doppler” isn’t an answer at all. It’s a term that you offer without understanding.

Radar and LIDAR haven’t yet worked successfully in self driving cars, so you think they’re going to work when you add another dimension and quadruple the speed?

Sure…

And you actually think that “yet undeveloped technology” will be created, developed, and tested in the next few years, to enable your prediction on AI timeline?

Sometimes, it’s better to let people wonder if perhaps you’re ignorant on a topic, than to keep posting and remove all doubt.

 

[MolaKule]

Isn't that the whole premise of AI. That it learns from billions of data points to become smarter and more capable than any human could ever be?

I think you may misunderstand a few points about AI.

Philosophically, no created entity can ever achieve a higher intelligence than its creator.

Artificial Intelligence, AI, (or Machine Intelligence) is also known as 'synthetic intelligence’. This includes understanding human speech, interpreting complex information from videos or photos, playing strategic games (such as chess and GO) against human experts, self-driving cars, and super search engines.

Software executes what it is instructed to do—and only that. Even where it is able to ‘learn’, that is only because it has had that learning capacity programmed into it. The software will blindly follow the algorithm (a sequence of instructions for carrying out a task) given to it. This defines how the program would calculate the identifying information and how to compare it with a stored profile. Note that software and algorithms are human intelligence conveyed to a computer.

AI is all about programming. Software is comprehensively tested to catch errors and ensure proper operation, but yet internal errors often surface with unexpected results. The news unfortunately often reveals problems with different information systems, such as entire airports having been shut down, or NASA losing a space probe in 1999 due to an information system failure. The Internet abounds with many such examples.

Some AI researchers consider the emergence in machines of super intelligence, above that of humans, to be only a matter of time. However, no matter how advanced the industry becomes, it still boils down to programming by human intelligence.

AI doesn’t dream about retirement on a tropical island, because consciousness is not merely a function of matter!

 

[Astro14]

Yeah, so far, ChatGPT has not figured this all out…

The reason this airplane has a buckled fuselage is that the nose wheel came down too hard. AI wants to LAND it on the nosewheel. Billions of calculations per second are meaningless without the contextual understanding of what AI is calculating. It leads to horribly wrong answers, like this one.

Chat, GPT by the way, only predicts what should be said next in a sentence. It does not understand context, or grammar. It is merely a language predictor. So it answers the above question without any contextual understanding of what a nose landing gear is, it merely predicts what should follow next in the sentence.

This is one of the problems with machine learning. It can go off on a wild, horribly wrong, tangent.

So when the machine sees something in an airplane that it’s never seen before, good luck to those in back…

 

[wpod]

Yeah, so far, ChatGPT has not figured this all out…

View attachment 170911
The reason this airplane has a buckled fuselage is that the nose wheel came down too hard. AI wants to LAND it on the nosewheel. Billions of calculations per second are meaningless without the contextual understanding of what AI is calculating. It leads to horribly wrong answers, like this one.

Chat, GPT by the way, only predicts what should be said next in a sentence. It does not understand context, or grammar. It is merely a language predictor. So it answers the above question without any contextual understanding of what a nose landing gear is, it merely predicts what should follow next in the sentence.

This is one of the problems with machine learning. It can go off on a wild, horribly wrong, tangent.

So when the machine sees something in an airplane that it’s never seen before, good luck to those in back…

Which returns to my post of. I would prefer a pilot driving the plane,,, a military pilot if possible.

 

[wpod]

After flying for lots of companies, 40 years and 25,000 hours , it doesn’t matter where a person got their “wings”.

What I know for a fact is, there are strong, average and a few weak pilots from all backgrounds.

It all comes down to the pilot.

I will 100% agree but Military pilots are put in to some nasty stuff. Years ago I had a continuing education class and there was a coastie pilot in the class. I asked him why do the Army Helos crash more often than any other branches of the military? He said respectfully the Army people are by nature of their mission parameters required to do more risky operations. That doesn't necessarily mean a better pilot , what that means is there may [or may not ] be a better chance of saving the day and my statement is not to slight a non military pilots skill or ability even though one of my dear friends is an Naval Aviator , retired and a major Airline Captain retired. He would kick my behind with extreme prejudice if I didn't say military pilots are the best pilots and Naval Aviators are the best of the best ,,, of the best.. Seriously kick my you know what !!!

 

[atikovi]

And you actually think that “yet undeveloped technology” will be created, developed, and tested in the next few years, to enable your prediction on AI timeline?

When Kennedy said in 1960 that man will walk on the moon within 10 years, how many people thought that prediction was crazy?

 

[Just a civilian pilot]

I will 100% agree but Military pilots are put in to some nasty stuff. Years ago I had a continuing education class and there was a coastie pilot in the class. I asked him why do the Army Helos crash more often than any other branches of the military? He said respectfully the Army people are by nature of their mission parameters required to do more risky operations. That doesn't necessarily mean a better pilot , what that means is there may [or may not ] be a better chance of saving the day and my statement is not to slight a non military pilots skill or ability even though one of my dear friends is an Naval Aviator , retired and a major Airline Captain retired. He would kick my behind with extreme prejudice if I didn't say military pilots are the best pilots and Naval Aviators are the best of the best ,,, of the best.. Seriously kick my you know what !!!