Will computers replace pilots?

seagull said:
You miss the point on the article entirely as the premise is based on not understanding the problem. This is an industry driven by data rates.
Click to expand...
No the article begins with a conclusion and builds to it. It makes the false assumption that the system can be designed to dynamically act (gangs in a standoff).

UAS don't "make decisions", they respond as programmed. If situation A exists proceed to response B and so on. None of the systems I ever flew had the latest version of Skynet we see on tv.

A single pilot cockpit will happen and the author points out why without even realizing it. The ability to free up tasks the pilot would otherwise perform will be used to enhance the efficiency of both the pilot and the operator.

Dispatch will be able to plan the flight based upon established routing by ATC. The aircraft will make use of their own separation logics to maintain lateral, vertical, and horizontal separation when in the auto-flight mode giving both ATC and the pilot "extra brain power" to combat other issues like mechanical failures and weather.

Imagine departures timed down to the millisecond. It's going to save a crap ton of time and energy, namely gas. That means money and if you're not aware, accountants run the world.

The pilot won't be removed but his job will change, it's already on the way. We've all heard about the children of the magenta who live and die by automation, it's happening right now. We are being trained and conditioned to accept autonomous systems as run of the mill normal.

Stick and rudder flying is going bye-bye right in front of us. I think it's a terrible idea but it is exactly the reality we have allowed to happen.
 
esa17 said:
The premise is only flawed if you have a superficial understand of the topic.

I don't, I have years of training as both a manned and unmanned pilot & mechanic. The tech is available and reliable. At this point the only barrier to entry is cost.

Articles like these have to be taken with a grain of salt. Just because I'm a pilot or stayed at a holiday inn doesn't mean I know anything about blimp safety. Watching the Hindenburg video only marginally increases my knowledge base.

Most folks who wade into the UAS debates find themselves in the same boat with Kevlar blinders on.
Click to expand...

As opposed to years of experience analyzing accidents and investigating them and knowledge of the systems you describe. No blinders, just reality based on experience.
 
esa17 said:
No the article begins with a conclusion and builds to it. It makes the false assumption that the system can be designed to dynamically act (gangs in a standoff).

UAS don't "make decisions", they respond as programmed. If situation A exists proceed to response B and so on. None of the systems I ever flew had the latest version of Skynet we see on tv.

A single pilot cockpit will happen and the author points out why without even realizing it. The ability to free up tasks the pilot would otherwise perform will be used to enhance the efficiency of both the pilot and the operator.

Dispatch will be able to plan the flight based upon established routing by ATC. The aircraft will make use of their own separation logics to maintain lateral, vertical, and horizontal separation when in the auto-flight mode giving both ATC and the pilot "extra brain power" to combat other issues like mechanical failures and weather.

Imagine departures timed down to the millisecond. It's going to save a crap ton of time and energy, namely gas. That means money and if you're not aware, accountants run the world.

The pilot won't be removed but his job will change, it's already on the way. We've all heard about the children of the magenta who live and die by automation, it's happening right now. We are being trained and conditioned to accept autonomous systems as run of the mill normal.

Stick and rudder flying is going bye-bye right in front of us. I think it's a terrible idea but it is exactly the reality we have allowed to happen.
Click to expand...

This misunderstands the reasons why it will not happen. Sure, there is apparent savings but you are operating from a fundamental misunderstanding of how and why accidents happen and what will prevent them. You are not unique to this, but you do not appear to understand the limitations. What is interesting is that you just proved yourself that your own premise is incorrect, do you see how you did that?
 
Now you're just utilizing a circular argument.

You're wrong because you're wrongy wronger wrong.

I didn't say a thing about reducing pilot error, I'm referencing the reduction in cost.

Accidents happen because of a chain of events, not because of magic bullets or golden BBs. Risk can be mitigated, not eliminated. Automation mitigates a great deal of risk but the unintended consequence is that other risks are introduced.

Training, procedure and consistency are the tried and true ways to mitigate risk but they sure cost a lot of money.
 
The power plant I worked at was for the most part automated, and could run itself. But only through steady state. It needed human intervention for starting and stopping, and whenever there was an upset. Automation is great, but for risky stuff like power plants and airplanes, there will always be the need for total human oversight.

People are stupid. Machines are designed by people, so they're inherently stupid too. No clear winner.
 
seagull said:
As opposed to years of experience analyzing accidents and investigating them and knowledge of the systems you describe. No blinders, just reality based on experience.
Click to expand...

Again a flawed assumption.

Just because one is skilled at analyzing the history of the English language doesn't mean they might be skilled at programming in C#.

They're both languages with structures and grammatical rules but they're worlds apart.
 
One day, yes. Also, the computers will fly better than any human pilot ever could. My bet is our kids will have to deal with the beginnings of unmanned passenger operations and our grandkids will definitely have to deal with it. My bet is it will coincide heavily with the advent of true AI, in which case unmanned passenger ops will be one of the least important aspects of our lives that are changed.

The amount of "it will NEVER happen" I hear from pilots is disheartening.
 
You "OMG it's coming right nao" people don't have any clue how many new failure modes will be discovered if this is ever introduced at scale.

We are not ready for this yet. The referenced article is spot on, though its presentation could be better.

People pushing back against autonomous, remotely-operated or single-pilot airliners aren't only fighting to keep their jobs—they're often pushing back against the concept strictly in the name of aviation safety.

-Fox
 
No, they are troglodytes.

The article and its supporters are speaking as if they are experts on something they have no understanding of. None.

You can hope that someone is like me is wrong, you can drag you feet as much as you want but I'm not. Jobs aren't going away, our career paths are just evolving.
 
esa17 said:
No, they are troglodytes.

The article and its supporters are speaking as if they are experts on something they have no understanding of. None.

You can hope that someone is like me is wrong, you can drag you feet as much as you want but I'm not. Jobs aren't going away, our career paths are just evolving.
Click to expand...

I do not accept your assertion that you are in a position to say who is unqualified to comment on these issues.

People who share the ideas that you espouse will eventually kill hundreds or thousands of people.

It's not that we can't, but that we can't safely, at scale. That we shouldn't.

It is not yet time for this.

-Fox
 
Acrofox said:
I do not accept your assertion that you are in a position to say who is unqualified to comment on these issues.

People who share the ideas that you espouse will eventually kill hundreds or thousands of people.

It's not that we can't, but that we can't safely, at scale. That we shouldn't.

It is not yet time for this.

-Fox
Click to expand...

Of all the people in this thread, Jeff is probably the most qualified to talk about this stuff man.
 
Remember the Qantas A380 that had the uncontained engine failure that punctured so many wires that the computer said "landing not possible"? Yet everyone walked away. Flying is not always black and white, and I'm positive that even IF we get a super computer that supposedly can fly autonomously and solves any emergency that it can face, during the first years of its life it will run into a problem IT won't be able to solve.

Heck, even in programming. A Murphy's law on programming states that if you create a piece of software that is so easy/good and fool proof that anyone can use, the first user will be the a fool that manages to actually make it crash.

I'm just saying....
 
ppragman said:
Of all the people in this thread, Jeff is probably the most qualified to talk about this stuff man.
Click to expand...

I wasn't questioning his qualification to talk about "this stuff".

"This stuff" spans a tremendous cross-section of knowledge domains, however, and every perspective is going to come with inherent bias. Working knowledge in one or two of those knowledge domains does not allow one to dismiss someone's statements out of hand by declaring them unfit to comment, or referring to anyone who disagrees with you as a 'troglodyte' or 'luddite'. Even experts who possess solid understandings of many facets of a complex problem will often strongly disagree with one another on what conclusions should be drawn from the available data.

I do not accept anyone on this forum as an authority on "this stuff" (nor do I profess to be), but I have lost track of how many times very smart people developing a technology, possessing intimate knowledge of it and a solid plan to implement, have started to work on a "simple" problem on scale, only to blow through their estimates by days, then weeks, then months, then sometimes years because of the operational complexity that comes with the implementation... edge cases, corner cases, issues requiring redesign bugs, and infinite black swans... and these are not safety-critical hundreds-of-lives-at-risk sorts of problems.

All of these problems are very well known by people who build large-scale systems. Proof-of-concept is very important, but it's an artificially-low bar, and it's not itself evidence that something can or should be implemented.

-Fox
 
esa17 said:
No, they are troglodytes.

The article and its supporters are speaking as if they are experts on something they have no understanding of. None.

You can hope that someone is like me is wrong, you can drag you feet as much as you want but I'm not. Jobs aren't going away, our career paths are just evolving.
Click to expand...
I'm interested in your thoughts on this. Let's for example look at a 30 year old flying for a major airline. How do you see his career in 10, 20, and 30 years from now?
 
I saw a concept of a google car-like drone. I think if that ever seriously gets off the ground and proves a reliable level of automation (AKA no one dies),then we might be one step closer. I highly doubt seeing pilotless/single pilot airliners happening on a timeline that involves my generation. The safety concerns are numerous. There are just some situations that takes a human to discern...

@MaRiO FDZ brings up the uncontained engine failure. Well I got a few other ones for you. How about that Kalitta 747 that split in two and aborted their takeoff. What about windshear/microburst events on an approach. What about crazy turbulence followed by heavy crosswind landings. We tout high levels of automation, but there are some serious issues that have not been solved or have not been solved to a safe degree yet. Don't aircraft with auto land systems have max crosswind limitations? There are a lot of outside factors requiring sophisticated indicating equipment with artificial intelligence requirements to react to those indications in a safe and timely manner that just don't make it safe or feasible for decades. Single pilot operation. I highly doubt it. Coordination involved takes a very busy two person crew environment making complex decisions that factor in many more variables than a computer can do. You show me a plane that's in imminent danger (dual engine failure, or fire on board) that necessitates an emergency landing ASAP that can find a place to land at Joe Bob's 5,000' runway or on terra firma on Billy Bob's flat farm land 5 miles west of the mountains. Sure we talk about how reliable planes are, and how when things are working 99.9% of the time it's all good, but when the excrement hits the proverbial fan the limits of computers is quickly found. If somehow there is a plane full of live people willing to risk their lives on a pilotless plane there will eventually be an "oh oops never thought about that" moment, lives will be lost, and it'll be back to the drawing board.

The complexities are enormous. Good on the people whose job it is to show us we're all wrong and that pilotless planes are the future, but it's a complex issue that will consume engineers and computer scientists for life times I think. I just don't see it happening in the 121 world. If that day comes in my life time, I'll gladly eat my words, but I'll be a optimist, and say that I probably will never see that day.

Then as I've said before on these types of threads. It's cost prohibitive both by design and red tape. It'll take a long time for that to change.

TL;DR Airline pilots are protected for a while as far as I'm concerned.
 
Last edited:
esa17 said:
I didn't say a thing about reducing pilot error, I'm referencing the reduction in cost.

Accidents happen because of a chain of events, not because of magic bullets or golden BBs. Risk can be mitigated, not eliminated. Automation mitigates a great deal of risk but the unintended consequence is that other risks are introduced.

Training, procedure and consistency are the tried and true ways to mitigate risk but they sure cost a lot of money.
Click to expand...

First, a reduction in cost is possible, but not likely. If it is an RPV you still need to pay someone on the ground and the crew cost is actually such a small percentage of flying airplanes that the reduction is scant. Assuming the best-case scenario I'll grant you there might be some reduction but is it enough to overcome the risk.

You are flat out WRONG that "accidents happen because of a chain of events" (at least when we look at modern aircraft accidents), that reflects thinking and accident models that were invented in the early part of the last century and have been shown to be only applicable to very simple systems. There is a science to that you are clearly not familiar with. I point you to reading up on resilience engineering. I think it is these simplistic models of events that lead you to the flawed assumptions about what we can automate with reasonable safety rates.

You are, therefore, also incorrect in your assumption on how to mitigate risk and that is a topic that is too long for me to bother with educating you on here.

Of course the article starts with a premise and then argues to support it, what article doesn't do that? The point is that in order to actually see further improvement it would require a dynamic system and just merely following a program leaves a lot of exposure to a variety of risks than are leading to aircraft accidents today.

Finally, while your knowledge as someone that is an operator and a mechanic is noteworthy, I will take mine from the control system engineers that are actually designing these for the major OEMs, thank you very much. As previously stated, we will reach a point that we will be able to turn all flight over to computers but not prior to having them reach the AGI level, or be extremely close to that.
 
Last edited:
I'm of the impression that there are two bottle necks at this time to this stuff "taking off".

1) The ATC system
2) Mixing computer controlled aircraft with those controlled manually

On one, the system is evolving right now with "NextGen" and the such. The ADS-B requirement will help. It's going to take a huge push by industry and government to nail down the specifics on operating autonomously in the environment.

On two, right now technology exists where aircraft in effect are in "super cruise" similar to cars. They use ADS-B and the autopilot maintains a set distance from the aircraft in front of it. Essentially you are providing your own spacing requirement similar to what ATC does these days. We're going to get to the point where a massive computer somewhere crunches numbers and spits out routing instructions and essentially we have to arrive at an approach gate at a specific time. Once this stuff starts happening single pilot aircraft aren't far behind.

Think about back in the 60's/70's, the Flight Engineer was an absolute requirement and arguably the busiest person in the cockpit. I highly doubt anyone back in those days could see systems being so automatic (basic by todays smartphone standards) that his job of transferring power, maintaining pressurization, fuel flow, etc. would go away. Systems now have completely taken all that logic off his hands, and so too will automation take out the second pilot.

Pretty much the only thing we have on our side is that the FAA moves at the pace of molasses. We'll be OK for 10-15 years. IMO. Beyond that I foresee single pilot airliners with the next generation beyond the 787/A350 being designed for single pilot operations.
 
I'd like to see a computer do the Hudson River landing.

But realistically, will there ever be pilot less passenger ops? Probably yes, but as has been pointed out, it will probably be an issue dealt with by our grandkids rather than ourselves. But there are many technological and ethical questions that need to be answered first, like the one the self driving cars are running into where if the car has to decide between running into a group of pedestrians or most likely killing the vehicles occupants, which does it choose? Same can be said with an aircraft, between flying over a populated area during an emergency or flying around it and increasing the chance of it crashing. This is why Google and other companies working on self driving cars are hiring all those unemployed philosophy majors to help with programming "morality and ethics decisions"
 
You must log in or register to reply here.
Back
Top