Will robots/AI replace human pilots?

Technology has reached its capacity. It would be impossible to assume that a computer could ever analyze an engine failure and then provide the best course of action based on preset or continuously updated variables. Never gonna happen.
 
Checkout_my_Six said:
Meh....it doesn't change that much. My answer sez....it dun matter.
Click to expand...
It changes enough that the 737NGs that I fly can rarely fly the VNAV profile that the FMC calculates even though it has all the parameters, including uplinked wind data at four altitudes, without pilot intervention on the descent.

Add that uncertainty to the short runway at LGA and trying for LGA would have been a bad choice even with the benefit of 20/20 hindsight.

You have to design an automated system which can work all this out before any of its programmers have the benefit of the data in the accident report. Current technology isn't even close.

Meanwhile, all trends in the industry are going in the opposite direction. We are improving safety by increasing the coordination between the pilots and the pilots and automation.
 
Velocity173 said:
Technology has reached its capacity. It would be impossible to assume that a computer could ever analyze an engine failure and then provide the best course of action based on preset or continuously updated variables. Never gonna happen.
Click to expand...

A few examples of where computing is already and will be commonplace in twenty years:

Personal assistants responding to natural language with 100 IQ or better intelligence. These assistance will be virtual clones. They will know you better than you know yourself, track your biological health and have your complete medical history. They will give advise and organize as much of your life as you allow.

Ubiquitous computing — also known as "pervasive computing" will be embedded everywhere and in everything including your body. These systems will also be intelligent and be performing tasks automatically. AI is already at the point where small machines will perform tasks now done by biological creatures such as bees, birds, domestic and commercial animals. Machines will manufacture artificial meat, organs for transplantation and transform atmospheric pollution. Nanomachines will reverse aging and act as hive problem solvers in agriculture and engineering (hyper intelligent buildings and vehicles).

Autonomous robotic intelligence (full singularity) will replace humans on the battlefield and in commerce (half of professions today will be automated including MD, CEOs, Engineering, Aviation, Architecture, the Arts, manual labor etc).
 
Velocity173 said:
Technology has reached its capacity. It would be impossible to assume that a computer could ever analyze an engine failure and then provide the best course of action based on preset or continuously updated variables. Never gonna happen.
Click to expand...
I didn't think about it that way. All valid points. Current technology is the pinnacle of all technology forever. Settle in. It's going to be a long boring ride from here. Lol
 
MD11Pilot said:
In addendum to the last post.
Boeing was clueless also because "it can't happen".
Click to expand...
"The difference between something that cannot possibly go wrong and something that might go wrong is when something that cannot possibly go wrong goes wrong, it's usually almost impossible to get at or fix."
- Douglas Adams
 
Velocity173 said:
Technology has reached its capacity. It would be impossible to assume that a computer could ever analyze an engine failure and then provide the best course of action based on preset or continuously updated variables. Never gonna happen.
Click to expand...

Nobody ever said that.

What someone did say was that an air traffic controller trying to pontificate on engineering makes as much sense as an engineer trying to pontificate on air traffic control. And yes, I've had a hand in some of your tools (some years ago, and specifically TMA and FAST), but I still wouldn't presume to know everything -- or even all that much -- about your job.

Most of what you said displays a critically superficial understanding that will positively lead YOU (no, not everyone) to a wrong answer.

There most definitely is a limit to what can or should be automated. You have stepped over that limit, without even understanding just how you stepped over it. That's well short of saying technology is done.
 
MAKG1 said:
Nobody ever said that.

What someone did say was that an air traffic controller trying to pontificate on engineering makes as much sense as an engineer trying to pontificate on air traffic control. And yes, I've had a hand in some of your tools (some years ago, and specifically TMA and FAST), but I still wouldn't presume to know everything -- or even all that much -- about your job.

Most of what you said displays a critically superficial understanding that will positively lead YOU (no, not everyone) to a wrong answer.

There most definitely is a limit to what can or should be automated. You have stepped over that limit, without even understanding just how you stepped over it. That's well short of saying technology is done.
Click to expand...

This thread is about the future. Key word "will." I don't think it takes an engineer to theorize what might be.
 
Velocity173 said:
This thread is about the future. Key word "will." I don't think it takes an engineer to theorize what might be.
Click to expand...
Yes, it does.

Otherwise, it bears absolutely no relation to reality.

Change your "will" to "might, if reality changes" and you'll be closer.

Having an AI choose a reliable emergency landing spot isn't impossible, but it's not on the horizon, either. What's on the horizon is what's close to being possible with existing technology. This doesn't qualify.
 
citizen5000 said:
A few examples of where computing is already and will be commonplace in twenty years:
Click to expand...
YouTube is filled with examples of such predictions from the past decades. While they occasionally have some things right, the majority of what they predicted is laughable today. Yet you state your predictions with absolute confidence?

Velocity173 said:
This thread is about the future. Key word "will." I don't think it takes an engineer to theorize what might be.
Click to expand...
The problem with this thread is that nobody defines their timeframe. For me, I'm talking about the next four or five decades--about twice the life of a brand new airliner. I don't see it happening for at least a century. I have no idea what will happen over several centuries. Many seem to suggest that the automated airliner is in our foreseeable future. I don't agree.
 
Velocity173 said:
I think that a robot could analyze the parameters of a dual engine failure in a split second and would have turned back to LGA and saved the entire crew, passengers and aircraft.:eek:
Click to expand...
Not in a million, or at least some dozen more, years - the sensor array to supply the "robot" with the parameters to make the call on such an out-of-the-norm situation aren't extant yet; It'd waffle about with re-start, not having "seen" the flock of boids heading in the intakes, not be smart enough to interface quickly enough with ATC and/or other traffic and "approve" it's own turn-back, likely getting muddles over the minimum separation parms, etc., and/or end up fighting with itself over the limits on configuration changes, mucking about understanding how the wind changes nearer to the surface, and a bunch of other stuff I didn't think of. Suffice it to say that level of integration, suitable for making that kind of decision, are not here. . .
 
MAKG1 said:
Yes, it does.

Otherwise, it bears absolutely no relation to reality.

Change your "will" to "might, if reality changes" and you'll be closer.

Having an AI choose a reliable emergency landing spot isn't impossible, but it's not on the horizon, either. What's on the horizon is what's close to being possible with existing technology. This doesn't qualify.
Click to expand...

Driverless cars have to make lots of on the fly decisions. They seem to be doing a pretty good job. Granted they are dealing with 2 dimensions, but it's a proof of concept.

If your position is specifically that airplanes will never at any point be able to autonomously pick a emergency landing spot sufficient to be beyond reproach.... you may be on to something.

If you can program a game airplane to be able to pretend it has pilot on board, then you already have a framework that shows the programming is possible. It's a matter of sensors and processing that input on a one to one basis for the hardware. You look at the data presented by one of the black boxes, and you see there are MANY sensors already in place to detect various factors. If a commercial pilot currently wants to, they can do the majority of tasks by telling the computer to execute their commands. The only thing missing at this point is auto-taxi.

Does that mean that I think pilotless planes are going to be a thing that happens tomorrow? No. But if driverless cars become a thing, watch for autonomous technology to expand rapidly.
 
MAKG1 said:
Yes, it does.

Otherwise, it bears absolutely no relation to reality.

Change your "will" to "might, if reality changes" and you'll be closer.

Having an AI choose a reliable emergency landing spot isn't impossible, but it's not on the horizon, either. What's on the horizon is what's close to being possible with existing technology. This doesn't qualify.
Click to expand...

You don't believe that something like Xavion (Gravity) could possibly be developed further to be certified in aircraft?

http://xavion.com/
 
ircphoenix said:
Driverless cars have to make lots of on the fly decisions. They seem to be doing a pretty good job. Granted they are dealing with 2 dimensions, but it's a proof of concept.

If your position is specifically that airplanes will never at any point be able to autonomously pick a emergency landing spot sufficient to be beyond reproach.... you may be on to something.

If you can program a game airplane to be able to pretend it has pilot on board, then you already have a framework that shows the programming is possible. It's a matter of sensors and processing that input on a one to one basis for the hardware. You look at the data presented by one of the black boxes, and you see there are MANY sensors already in place to detect various factors. If a commercial pilot currently wants to, they can do the majority of tasks by telling the computer to execute their commands. The only thing missing at this point is auto-taxi.

Does that mean that I think pilotless planes are going to be a thing that happens tomorrow? No. But if driverless cars become a thing, watch for autonomous technology to expand rapidly.
Click to expand...

I observe driverless cars on a regular basis. I get cut off by them all the time. I drive behind them doing 20 MPH in a 35 zone and blocking traffic behind them. Occasionally, they screech to a halt when two pedestrians stop on a street corner to talk. They don't seem to work in the rain, and their accident rate is consistent with the general population (it's not zero, and they haven't driven enough miles to establish reasonable statistics).

I thoroughly disagree that that is "pretty good." I classify a driver like that as incompetent. It's not ready, and it's not close.

Their behaviors are still not completely specified. They won't be ready for prime time without that. Study the "trolley problem" for an example of what that particular problem is.

And those are the Google and Nissan cars that have been at it for a while. The Uber cars are a LOT worse.
 
Last edited:
Velocity173 said:
You don't believe that something like Xavion (Gravity) could possibly be developed further to be certified in aircraft?

http://xavion.com/
Click to expand...
Umm, that only works for airports. Sure, you could certify that. But you had better not fly out of gliding distance of an airport if you want it to be even slightly meaningful.

Or fly in real wind that varies in time and space
 
just wait till drones start using that technology.....:ohsnap:
MAKG1 said:
I observe driverless cars on a regular basis. I get cut off by them all the time. I drive behind them doing 20 MPH in a 35 zone and blocking traffic behind them. Occasionally, they screech to a halt when two pedestrians stop on a street corner to talk. They don't seem to work in the rain, and their accident rate is consistent with the general population (it's not zero, and they haven't driven enough miles to establish reasonable statistics).

I thoroughly disagree that that is "pretty good." I classify a driver like that as incompetent. It's not ready, and it's not close.

Their behaviors are still not completely specified. They won't be ready for prime time without that. Study the "trolley problem" for an example of what that particular problem is.

And those are the Google and Nissan cars that have been at it for a while. The Uber cars are a LOT worse.
Click to expand...
 
I don't know much, but how do pilots of even today's airliners with autopilot get checked on maintaining proficiency?

I saw some disturbing (if their conclusions we true) accidents explained in aircraft investigations, where some pilots, when they had to take over were not proficient. One was at SFO.

And from here, I get the impression that some airline or commercial pilots see it as just a job, while others lov flying and still keep flying GA or other as well, to keep up their chops.

But it's all conjecture for me. No real knowledge of how it all hangs together.
 
MAKG1 said:
I observe driverless cars on a regular basis. I get cut off by them all the time. I drive behind them doing 20 MPH in a 35 zone and blocking traffic behind them. Occasionally, they screech to a halt when two pedestrians stop on a street corner to talk. They don't seem to work in the rain, and their accident rate is consistent with the general population (it's not zero, and they haven't driven enough miles to establish reasonable statistics).

I thoroughly disagree that that is "pretty good." I classify a driver like that as incompetent. It's not ready, and it's not close.

Their behaviors are still not completely specified. They won't be ready for prime time without that. Study the "trolley problem" for an example of what that particular problem is.

And those are the Google and Nissan cars that have been at it for a while. The Uber cars are a LOT worse.
Click to expand...

I've followed behind worse humans than I have followed behind terrible AI cars.

And the majority of accidents are when the human is operating the car. The first actual collision of a driverless car was a low speed side swipe of the bus in a construction zone. I'd say the development of the cars has proceeded by leaps and bounds in the last decade. The next decade should be even more incredible.
 
citizen5000 said:
The simulations showed the plane could have landed but the scenario failed when adding in the delay time Sully had when he made the decision to fly out and land on the river. A machine would have made those calcs quicker and landed back on the runway.
Click to expand...

I find it quite interesting that you can state for certain that an algorithm that has not even been developed would have come to a particular resolution. Your crystal ball is quite amazing.
 
I can understand a 737 but a UH-1? Nope, that's fake news.
 
The one thing about the robot landing the 737... so it was highly dependent on autoland... and really the robot was entirely unnecessary to do any of those things. If you incorporate servos into the controls that don't have them already, you can do away with the need for the robot entirely. Very interesting proof of concept though.
 
ircphoenix said:
The one thing about the robot landing the 737... so it was highly dependent on autoland... and really the robot was entirely unnecessary to do any of those things. If you incorporate servos into the controls that don't have them already, you can do away with the need for the robot entirely. Very interesting proof of concept though.
Click to expand...
for now....that bot makes the aircraft a dual use vehicle and eases the workload for single pilot operations that can later morph as the technology matures to a totally autonomous operation. Note a new vehicle design is not needed.....:eek: But ,future vehicles will come with built in servos and remote digital controls vs hand knobs. The bot will be more integrated or designed out....as future designs may not require as many human interfaces.
 
citizen5000 said:
IBM’s Watson (AI Medical System) Delivers Proper Diagnosis for Japanese Leukemia Patient In 10 Minutes After Doctors Failed For Five Years http://www.nydailynews.com/news/world/ibm-watson-proper-diagnosis-doctors-stumped-article-1.2741857
Click to expand...
A single dimensional problem with a minimal number of logical paths. I don't know what the multiplier is even if it is a linear relationship, but everything that Sully decided and effected involved many more logical paths in much less time. additionally this doesn't even consider the practice and feel involved ingliding a plane successfully to a zone and setting it down at the best speed and pitch attitude for the situation. I have been assuming through all this that I am discussing this with a pilot, but I am beginning to wonder.
 
MBDiagMan said:
Everything that Sully decided and effected involved many more logical paths in much less time.
Click to expand...

Let's backup a bit.

Early gaming systems used a network of nodes (data points) arranged in a tree (these algorithms were first described back in the 1950s). Machines were unable to best a human grand-master until 18 years ago when IBM's Deep Blue attained processing speeds fast enough to beat Kasparov. Still using the algorithms defined from the early days of mainframes. Today, handheld computers can beat any human playing chess or any other game you choose.

Medical AI in infancy. Given the constraints of processing speeds and the linear nature of rule-based systems complex medical problem strategies have been developed to limit the number of hypotheses that a program must consider and to incorporate pathophysiologic reasoning. The latter innovation permits a program to analyze cases in which one disorder influences the presentation of another. AI in medicine must for now depend on preprocessing what the machine deals with by organized around models of disease. Here is a white paper from MIT on the subject: http://groups.csail.mit.edu/medg/ftp/psz/SchwartzAnnals.html

Machine speed is key to advancing not only AI but every other field now largely dependent on computing solutions. Soon today's machines (despite being thousands of times faster than even 18 years ago) will be dwarfed by quantum processors and quantum memory and nanotechnology. Eventually medical systems will likewise be able to diagnose without preprocessing and when combined with improved sensors humans will become bystanders as the machines both diagnose and treat disease.

Sully's decision "tree" was tiny compared even to a simple checkers playing program. A machine would instantly compute all possible solution paths in nanoseconds (not whole seconds) and turned the plane to the field and landed it.
 
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 am still curious as to whether or not you are a pilot, and if so, at what level.

BTW, we are talking about flying. This is a life and death activity, unlike your video games you seem to be fixated on. The medical AI is also much different. When you get to medical AI that involves complex, experience required surgery, then we can maybe equate the actual surgery manipulation with landing an airplane in adverse conditions.
 
Checkout_my_Six said:
I wonder how smooth auto land is.....vs. the old man up front?
Click to expand...

I've only flown one plane that autolands (two others were technically capable, but the company didn't spend the money to keep them certified), and sometimes it did a better job, and sometimes not. It didn't handle winds well at all, and crosswinds near the limit (15 knots) were sporty. That said, stronger winds and low visibility (when you'd do an autoland) don't usually go together.
 
Last edited:
LongRoadBob said:
I don't know much, but how do pilots of even today's airliners with autopilot get checked on maintaining proficiency?

I saw some disturbing (if their conclusions we true) accidents explained in aircraft investigations, where some pilots, when they had to take over were not proficient. One was at SFO.

And from here, I get the impression that some airline or commercial pilots see it as just a job, while others lov flying and still keep flying GA or other as well, to keep up their chops.

But it's all conjecture for me. No real knowledge of how it all hangs together.
Click to expand...
We're in the sim every 6,8 or 12 months (depending on the company) plus the captain has to get evaluated on the line every 18 months at my company. The majority of GA pilots don't do any recurrent training. I'd take the pro pilot over the GA pilot almost all the time.
 
MBDiagMan said:
I am still curious as to whether or not you are a pilot, and if so, at what level.
Click to expand...

We've been asking him this for a long time, and his silence tells me everything I need to know. We're dealing with a Cirrus shill that gets everything he posts from a Google query. Nothing more.
 
Checkout_my_Six said:
I wonder how smooth auto land is.....vs. the old man up front?
Click to expand...
They're average, at best. The flare is done by rote. Everything is based on fixed radar altitudes; power-to-idle, pitch up, etc. That works well on a calm day--which is normal for the low visibility landings that the systems were designed for--but they don't handle crosswinds or gusts gracefully and it doesn't take much for the winds to exceed their limitations.
 
kayoh190 said:
We've been asking him this for a long time, and his silence tells me everything I need to know. We're dealing with a Cirrus shill that gets everything he posts from a Google query. Nothing more.
Click to expand...

Okay. Thanks! Once I looked back through some of his stuff, it becomes glaringly obvious that he is probably writing from Daddy's basement.
 
Checkout_my_Six said:
one step closer....:D

http://www.aero-news.net/getmorefro...post&id=982870f4-c9f1-4d82-8487-231b6a240f9d#
Click to expand...

Checkout_my_Six said:
so easy a robot can do it.....in the future, right?. :D

https://www.avweb.com/avwebflash/news/Robot-Co-pilot-Lands-737-229022-1.html
Click to expand...

Post #211.

And you posted the same story twice? LOL.

But seriously on that first link, shame on you for linking to that azzz-hat's website for a news source.

There's members here who were sued by that prick. Nobody should ever click any link to that site, ever, IMHO.
 
You must log in or register to reply here.
Back
Top