Are pilots going to be replaced by AI?

[Pinecone]

14 hours ago, Jerry 5TJ said:

Although it may take some time to implement and perfect such systems,

And how many crashes/deaths along the way?

I have worked in and around the computer industry a long time.  And complex systems have complex failure modes that are not easily anticipated.  And when you have a failure, the fix often causes other failures.

[Pinecone]

3 hours ago, ilovecornfields said:

One pilot on the ground could monitor several aircraft flown by AI. In fact, this is already happening in some cases. For example, the US Air Force is using drones that are flown by AI, and these drones are monitored by pilots on the ground. The pilots are able to monitor the drones and take control if necessary.

There are a number of reasons why it might be beneficial to have one pilot on the ground monitoring several aircraft flown by AI. For example, it would allow for greater efficiency, as the pilot could monitor multiple aircraft at once. It would also allow for greater flexibility, as the pilot could be located anywhere in the world.

However, there are also some challenges that would need to be overcome in order to make this a viable option. For example, the pilot would need to have a very good understanding of the AI systems that are being used to fly the aircraft. They would also need to be able to react quickly to any unexpected situations.

And what happens if more than one of the aircraft being monitored has some issue/failure?

[N201MKTurbo]

As @Pinecone stated, the more complex a software system gets, the harder the failure modes are to anticipate, and a fix applied to one system is unlikely to apply to all systems.

like all aviation regulations, they are usually written in blood. Only time will tell how much blood is spilled before we achieve an acceptable level of risk.

[N201MKTurbo]

One thing to keep in mind about this modern AI technology, it can’t easily be embedded in a stand alone system. It relays on huge databases (the internet) to function. This would mean that the communications channel to the internet would be necessary for the AI to function. That seems like a critical vulnerability to me.

[N201MKTurbo]

Do you think the Garmin auto land system could keep working and safely land the plane if the GPS system was jammed? I don’t think so. If an airplane looses all external navigation input, the PILOT can land the airplane visually. I don’t know of any automation system that can do that.

[Hank]

2 hours ago, N201MKTurbo said:

Do you think the Garmin auto land system could keep working and safely land the plane if the GPS system was jammed? I don’t think so. If an airplane looses all external navigation input, the PILOT can land the airplane visually. I don’t know of any automation system that can do that.

To say nothing of electrical issues on the plane. I've had a total electrical failure; when one component on the cockpit dimmer circuit board gave up, it took everything with it. Just after breakout on a VOR-A approach in West-by-Gawd, Virginny, rolling inbound over the VOR. 

I was able to stop the descent, evaluate my options, and fly home along the Ohio River ~2000msl with the gear down, and make a normal landing at my uncontrolled home drove. Ain't no AI gonna do that, its turning off with everything else, followed shortly by --SPLAT--!!

[ilovecornfields]

Dear readers,

I am writing to apologize for all of my posts that were made by an AI. I know that this may come as a surprise to some of you, and I want to assure you that I am taking full responsibility for my actions.

I started using an AI to write my posts because I was struggling to find the time to write as much as I wanted to. I thought that using an AI would help me to produce more content, but I now realize that it was a mistake.

I know that using an AI is considered cheating, and I apologize for any inconvenience it may have caused. I also understand that some of you may be disappointed in me, and I want to assure you that I am truly sorry for what I have done.

I have learned my lesson, and I will never use an AI to write my posts again. I will continue to write my own content, and I will do my best to make sure that it is of the highest quality.

Thank you for your understanding.

Sincerely,
[Your name]

[carusoam]

29 minutes ago, ilovecornfields said:

Dear readers,

I am writing to apologize for all of my posts that were made by an AI. I know that this may come as a surprise to some of you, and I want to assure you that I am taking full responsibility for my actions.

I started using an AI to write my posts because I was struggling to find the time to write as much as I wanted to. I thought that using an AI would help me to produce more content, but I now realize that it was a mistake.

I know that using an AI is considered cheating, and I apologize for any inconvenience it may have caused. I also understand that some of you may be disappointed in me, and I want to assure you that I am truly sorry for what I have done.

I have learned my lesson, and I will never use an AI to write my posts again. I will continue to write my own content, and I will do my best to make sure that it is of the highest quality.

Thank you for your understanding.

Sincerely,
[Your name]

Which AI software did you use… ChatGPT?

was that running on an NVDA H100 chip?

was that using a work related server?   (Work has got to be an interesting application/challenge for AI, FDA and FAA safety levels)

 

Best regards,

-a-

[N201MKTurbo]

The strange thing is, I remember having this same conversation with an air traffic controller back in the 80s. I was taking the side of automation. 40 years of flying and engineering experience has changed my viewpoint.

[carusoam]

Speaking of AI that didn’t go as planned…

Version 2 Starlink sats are off to a crummy start…. Some were getting de-orb’d after their recent launch… no details why just yet.

In other SpaceX news… the SpaceX Starship is expected to attempt launch on 4/20… spread the rumor….   

 

PP rumor mill only… no details to quote.

Best regards,

-a-

[N201MKTurbo]

I think in a world controlled by SkyNet, we will see different accidents than today. Only time will tell if there are less.

[N201MKTurbo]

3 minutes ago, carusoam said:

Speaking of AI that didn’t go as planned…

Version 2 Starlink sats are off to a crummy start…. Some were getting de-orb’d after their recent launch… no details why just yet.

In other SpaceX news… the SpaceX Starship is expected to attempt launch on 4/20… spread the rumor….   

 

PP rumor mill only… no details to quote.

Best regards,

-a-

An interesting choice of dates.

[N201MKTurbo]

5 minutes ago, 1980Mooney said:

But to be fair how many commercial aircraft routinely landed by AutoLand in 1980?  And how many routinely land by AutoLand today?  And how many commercial aircraft built this decade are capable of AutoLand? - isn’t it all of them? It’s in high end personal turboprops, small jets. Imagine what advances in capability is on the drawing board for late 2020’s commercial aircraft. 
 

The situation has significantly changed. 

I think we are conflating fancy autopilots with automated decision making. At this point, the pilot will still command the plane to auto land. I know that some systems will auto land if the pilot stops doing anything, but we are talking about remote control, turning all aircraft into drones. I can see it happening in the airlines first, but we won’t be able to afford the equipment. 
 

If this was to come about, can you conceive of somebody with no license or training being able to fly a general aviation airplane? After all, a pilot isn’t required any more.

[ilovecornfields]

28 minutes ago, carusoam said:

Which AI software did you use… ChatGPT?

was that running on an NVDA H100 chip?

was that using a work related server?   (Work has got to be an interesting application/challenge for AI, FDA and FAA safety levels)

 

Best regards,

-a-

I used Google Bard from my phone but I wasn’t the only AI on this thread. I wanted to try it as an experiment after I used Google Bard to write a response to the “French AI” guy in another thread and the AI bit ended up writing a much better response than I would have. I wanted to see how far it would go but someone sent me a PM this morning after they’re figured it out so I realized it was time to confess.

For the record, I don’t think AI should fully replace pilots but I predict single pilot+AI will be coming in the not too distant future.

[carusoam]

3 minutes ago, 1980Mooney said:

 

That’s a great idea. I need GPT-3 Chatbot to start monitoring Ang writing responses for me. I spent waaaay too much time here

Some of your posts are going to be extra challenging for AI to come up with…

Lots of flight data research involved…!   

 

I’d be happy with an improved search function around here…

AI could certainly help with that…

Best regards,

-a-

[Schllc]

14 minutes ago, 1980Mooney said:

But to be fair how many commercial aircraft routinely landed by AutoLand in 1980?  And how many routinely land by AutoLand today?  And how many commercial aircraft built this decade are capable of AutoLand? - isn’t it all of them? It’s in high end personal turboprops, small jets. Imagine what advances in capability is on the drawing board for late 2020’s commercial aircraft. 
 

The situation has significantly changed. 

All valid, however flying isn’t all about getting from point A to point B in GA. Neither is driving or boating. There is an element of the spirit of adventure that is uniquely human in all these activities.  All at a level AI will never be able to compute, much less comprehend. 
There will likely be a day where the person who actually wants to pilot an aircraft using their wits and skill is a tiny minority, but as long as the human element exists, so will that spirit. 

I have a good friend that I have done many adventure with over the years. 
We were at a gathering a few years ago and his wife was telling a story about a camper jackknife accident they had years ago. One of our friends commented, “Wow, stuff always happens to you guys”, she thought about it for a moment and said “it’s because we do shit!, not a lot of danger sitting on your couch at home!”.

Flying has risks, as does anything fun or adventurous. AI likely has its place in commercial activity and perhaps for those who prefer to rely on a computer rather than their own skills.  I sure hope the freedom to choose between the two is eternal. 

[N201MKTurbo]

3 minutes ago, ilovecornfields said:

I used Google Bard from my phone but I wasn’t the only AI on this thread. I wanted to try it as an experiment after I used Google Bard to write a response to the “French AI” guy in another thread and the AI bit ended up writing a much better response than I would have. I wanted to see how far it would go but someone sent me a PM this morning after they’re figured it out so I realized it was time to confess.

For the record, I don’t think AI should fully replace pilots but I predict single pilot+AI will be coming in the not too distant future.

So, when you have to pull the plug on HAL, will one pilot be able to handle the workload? Will the pilot be in command or HAL?

[whiskytango]

16 hours ago, MooneyMitch said:

HAL 9000 

Message you do not want to receive in your AI - enabled aircraft:

"Just a moment. Just a moment. I've just picked up a fault in the AE-35 unit. It's going to go 100-percent failure within 72 hours."

[N201MKTurbo]

What do you do when your AI airplane wakes up cranky some day and argues with you about everything?

[MooneyMitch]

7 minutes ago, N201MKTurbo said:

What do you do when your AI airplane wakes up cranky some day and argues with you about everything?

Come on man, we all have experience with this issue..................... 

[jlunseth]

Are you kidding? What is the Bill Gates solution to every computer problem, the solution that has existed since DOS? Its "reboot." Why is there such a thing? Because computers fail for inexplicable reasons. Do you really want to be a passenger in something that needs to reboot to fix itself? What is the aircraft going to do while the computer is deciding how to become a computer again? Humans make mistakes. Computers also make mistakes. But what computers do fairly often that humans do rarely is simply stop functioning.

[ilovecornfields]

24 minutes ago, N201MKTurbo said:

So, when you have to pull the plug on HAL, will one pilot be able to handle the workload? Will the pilot be in command or HAL?

Autopilot disconnect 

[carusoam]

Hal sends a request to the passengers in the back….

Uhhhh… Is there anyone back there with DOS experience, ya know what I mean…? 

-a-

[MooneyMitch]

AI replacing pilots............................SpaceX Crew Dragon flights to/from International Space Station...................no pilots required...........those astronauts are just along for the incredible adventure!! WOW!!!

With that said, I've never heard any discussion regarding the astronauts being able to, or being trained to override AI systems if need be.  Hmmm...............

[jlunseth]

52 minutes ago, 1980Mooney said:

Really? How many humans on Voyager 1 would still be functioning?  And that is a Stone Age computer. 

Sure. And its not in any danger of falling out of space whenever it needs to reboot. See e.g. https://space.stackexchange.com/questions/24041/how-is-voyager-1-still-operating . Voyager does not need to deal with the laws of aerodynamics and gravity.