Will computers replace pilots?

[seagull]

This is still an error, regardless of what we call it - yeah, pulling resilience out of a system is by and large "bad" and systems should be "engineered" to increase resilience to calamity...but ignoring a sink rate alert on approach right as you're breaking out of the clouds is an error. Yes, we "relied upon human performance to fill in the gaps there," but the bottom line is, the last "link in the chain" or "hole in the cheese" or "percolation theory" or whatever model you want to use to describe the accident in this accident was the pilot's decision to continue after the system telling him "GO AWAY FROM THE GROUND!" in no uncertain terms. Yes there are things that lead up to this accident (namely fatigue and confusion in the cockpit) but that's ultimately where conditioned responses to dangerous scenarios are key. Similarly, look at the Asiana accident - yes, I see how this could happen, and I'm certainly not immune, but these guys couldn't fly a visual approach. Yes I understand it's a complicated airplane, complicated airspace, and there were other cultural and linguistic issues that likely didn't help...but if you're not on speed on your approach... you go around - those guys didn't. Pilot error.

We are responsible for our own competency - no one else, not the company, not the FAA, not the engineers who designed the aircraft - us. We are responsible for our actions - it's not the result of a flawed system if you forget to drop the gear. It's not the fault of a flawed system if you CFIT - yes, better systems likely would make operations more resilient and damage tolerant, but it is not a panacea. If there are issues where we have to rely on human pilots to compensate for weaknesses in the system, then it is the responsibility of the human pilots (nay, their ethical duty) to assure that they are able to fill in the gaps.

You miss the fact that under stress people often do not hear warnings and that we have found many cases where humans never heard an audible alarm due to stress or fatigue. Is that an "error" on the part of the pilots, or is that an "error" on the part of people who designed the system who did not consult with actual human performance experts that understood that factor?

Asiana is similar. Eye scans of pilots flying approaches after a few years of autothrottle usage indicates that they drop the airspeed out of their scan, although they, themselves, are sure that they are monitoring it. In that case we have a pilot that had been flying narrowbody FBW Airbus products within Asia for a number of years. The Airbus architecture would not have allowed for the situation that autothrottles can be on yet they not respond (as a sidenote, one of the 787 engineering test pilots almost crashed due to the same issue on a 787, do you think he also "could not fly a visual approach?). There is a lot more, but much of these false ideas of human behavior come from ad hoc simplistic models that lay people put together, and yes, even though someone might be a pilot does not make them experts in all the nuances. So, again, do we have the "error" by the pilots or is the "error" in those that designed the systems and procedures? Do you think the pilots intentionally did not notice something?

 

[Soku39]

You miss the fact that under stress people often do not hear warnings and that we have found many cases where humans never heard an audible alarm due to stress or fatigue. Is that an "error" on the part of the pilots, or is that an "error" on the part of people who designed the system who did not consult with actual human performance experts that understood that factor?

Asiana is similar. Eye scans of pilots flying approaches after a few years of autothrottle usage indicates that they drop the airspeed out of their scan, although they, themselves, are sure that they are monitoring it. In that case we have a pilot that had been flying narrowbody FBW Airbus products within Asia for a number of years. The Airbus architecture would not have allowed for the situation that autothrottles can be on yet they not respond (as a sidenote, one of the 787 engineering test pilots almost crashed due to the same issue on a 787, do you think he also "could not fly a visual approach?). There is a lot more, but much of these false ideas of human behavior come from ad hoc simplistic models that lay people put together, and yes, even though someone might be a pilot does not make them experts in all the nuances. So, again, do we have the "error" by the pilots or is the "error" in those that designed the systems and procedures? Do you think the pilots intentionally did not notice something?

So your posit is that it was the planes fault that 4 777 pilots couldn't manage to safely get the plane on the ground in clear and a million conditions?

 

[seagull]

So your posit is that it was the planes fault that 4 777 pilots couldn't manage to safely get the plane on the ground in clear and a million conditions?

Far more complex than that and, like the AF447 pilots, these were not weak pilots. It is this type of response that prevents us from learning the lessons that would actually prevent another accident of a similar type. The fact that the pilot put it on a sight picture that was similar to his previous airplane along with did some procedures that would have had a totally different outcome in his previous aircraft also had some influence over the outcome. More to the point, do you really think you can prevent an "error" by telling someone to "be more careful" and "pay attention more"?

 

[ppragman]

So your posit is that it was the planes fault that 4 777 pilots couldn't manage to safely get the plane on the ground in clear and a million conditions?

That's exactly what he's saying.

 

[ppragman]

Far more complex than that and, like the AF447 pilots, these were not weak pilots. It is this type of response that prevents us from learning the lessons that would actually prevent another accident of a similar type. The fact that the pilot put it on a sight picture that was similar to his previous airplane along with did some procedures that would have had a totally different outcome in his previous aircraft also had some influence over the outcome. More to the point, do you really think you can prevent an "error" by telling someone to "be more careful" and "pay attention more"?

No, but you can certainly train people better so that they're more able to solve problems in the airplane in a manner that maximize the possibility of success.

 

[ppragman]

You miss the fact that under stress people often do not hear warnings and that we have found many cases where humans never heard an audible alarm due to stress or fatigue. Is that an "error" on the part of the pilots, or is that an "error" on the part of people who designed the system who did not consult with actual human performance experts that understood that factor?

Asiana is similar. Eye scans of pilots flying approaches after a few years of autothrottle usage indicates that they drop the airspeed out of their scan, although they, themselves, are sure that they are monitoring it. In that case we have a pilot that had been flying narrowbody FBW Airbus products within Asia for a number of years. The Airbus architecture would not have allowed for the situation that autothrottles can be on yet they not respond (as a sidenote, one of the 787 engineering test pilots almost crashed due to the same issue on a 787, do you think he also "could not fly a visual approach?). There is a lot more, but much of these false ideas of human behavior come from ad hoc simplistic models that lay people put together, and yes, even though someone might be a pilot does not make them experts in all the nuances. So, again, do we have the "error" by the pilots or is the "error" in those that designed the systems and procedures? Do you think the pilots intentionally did not notice something?

If he crashed almost crashed on a visual approach...then yes, he is weak at flying visual approaches, he should receive more training in them.

 

[MikeD]

Some items I've noticed here.

First off, whether we like it or not, it is the pilot's ultimate responsibility to ensure that he doesn't hit other aircraft, hit the ground, or hit anything attached to the ground, with his aircraft, that wasn't intended as so. To put the concept simply. Sometimes, pilots fail at that. What needs to be understood is that the phrase "pilot error" is not a cause(s) of an accident. What "pilot error" is, is a general descriptor of causal factor(s) of an accident. You will never see an NTSB or other accident report that states "the probable cause of this accident was pilot error. That is all"....with no description of what said error(s) was/were, no causal factors, no explanation, nothing to learn from, etc. The problem is, this term is used loosely by many both within aviation as well as outside of aviation, without asking the requisite questions: Ok, the causal factors description is "pilot error", what kind of error(s) occurred? When, how and why did they occur? Were they errors of commission, omission, or both? Were there any warning signs? Any mitigating factors or attempts to mitigate? How did said error(s) affect the situation in progress when they occurred? Etc, etc, etc. Those are just a few of many questions that need to be asked, when defining that "pilot error" was a cause descriptor of a particular accident.

It's no different than saying "mechanical failure" caused an accident. Sure, "mechanical failure" is the descriptor, but what kind of mechanical failure? What system? When/how/why? How did it affect? Was it singular, or part of a number of events? Were there cascading or related failures? Was it contributory to other failures of systems? How did it affect the situation at the time of occurrance? Etc, etc, etc. Same concept when saying "weather", or any other general descriptor.

"Pilot error" is not an incorrect term, however it is a term that has to used in the proper context, for what it truly is and means.

With regards to accident modeling, and things such as an "accident chain", or Jim Reason's "Swiss Cheese" model, or any number of other models of accident causation, it's not that the "accident chain" no longer applies or is wrong, it's been discovered that it's not the only methodology in which accidents occur. We have discovered new models over time and as the science/technology has evolved accordingly, we have further discovered that there is 'more than one way of doing business' when it comes to how accidents occur. There may still be accidents which happen to follow a "chain" model, especially if items/factors are interrelated to one another, and there will be accidents that follow other models, and even....there will be accidents that follow more than just one of the models either throughout the accident itself, or at different times of the overall accident situation.

It is all dependant on the dynamics of the particular accident, what model(s) they happen to follow, and when those model(s) apply. Granted, we're not finding many new ways to crash airplanes, very generally speaking (we are finding new ways when it comes to technology, however), yet even so, the nuts and bolts details of every accident is still unique to itself, hence why the specific dynamics of a particular accident is so important, and how different accidents can follow different models, and do.

So back to the general concept of "pilot error" when it comes to accident causal factors. The human will sadly still be one of both the strongest, as well as the weakest, points of failure in an accident, depending on the dynamics of the accident. We can institute all the safety features in the world, and still have pilots crashing planes for errors of commission or omission. There are any number of reasons this occurs, but for situations where it was an error made by the pilot/crew, we can't automatically exonerate the pilots 100%, anymore than we can automatically blame them 100%. Like has been described in this very thread, any number of factors that relate to the crew can/will be present. And only through looking at those and really doing the work of sorting the wheat from the chaff during the investigation, can we determine how much of the error made by the crew is either on their shoulders alone or which they share blame with something else.

 

[esa17]

If he crashed almost crashed on a visual approach...then yes, he is weak at flying visual approaches, he should receive more training in them.

I'm pretty sure private pilot lesson two includes "throw yourself at the ground and miss."

 

[MikeD]

You miss the fact that under stress people often do not hear warnings and that we have found many cases where humans never heard an audible alarm due to stress or fatigue. Is that an "error" on the part of the pilots, or is that an "error" on the part of people who designed the system who did not consult with actual human performance experts that understood that factor?

In the sense that ultimately, it's the job of the pilot to safely execute the flight, its their error. A secondary factor would indeed be "might we have a safety/warning system that isn't as effective as we believed it to be?"

If a pilot fails to lower the landing gear on an approach to a runway in high crosswinds, and he's so engrossed in mentally concentrating on the landing that he lands gear up, and only when the sliding stops does he notice a gear warning horn blaring at him, then the primary causal factor is indeed the failure of the pilot to lower the landing gear (and the related reasons of how/why that occurred), a secondary factor would be the warning system and did it work as advertised? Is it truly effective? Etc.

Because ultimately, it is the pilots responsibility to lower the landing gear for landing. Bottom line of PIC responsibility.

Asiana is similar. Eye scans of pilots flying approaches after a few years of autothrottle usage indicates that they drop the airspeed out of their scan, although they, themselves, are sure that they are monitoring it. In that case we have a pilot that had been flying narrowbody FBW Airbus products within Asia for a number of years. The Airbus architecture would not have allowed for the situation that autothrottles can be on yet they not respond (as a sidenote, one of the 787 engineering test pilots almost crashed due to the same issue on a 787, do you think he also "could not fly a visual approach?). There is a lot more, but much of these false ideas of human behavior come from ad hoc simplistic models that lay people put together, and yes, even though someone might be a pilot does not make them experts in all the nuances. So, again, do we have the "error" by the pilots or is the "error" in those that designed the systems and procedures? Do you think the pilots intentionally did not notice something?

I'd look at training. Your descripton of the above regards the autothrottles, reads to me that this is a known potential problem/issue with guys who transition from aircraft type X to aircraft type Y, or manufacturer type X to manufacturer type Y. If that is the case, then did the training program the crew received cover this extensively as a known "gotcha" that has bitten crews past? How much was this emphasized in ground, simulator, an/or course study materials?

The error isn't necessarily in the system, IF the pilot is either unfamiliar with the system, doesn't know its nuances, or doesn't utilize the system as its supposed to be used.

Ultimately, the crew made the error and crashed a perfectly good operating airplane. The why of it all, could reveal other issues such as the above.

 

[ppragman]

@MikeD

A Navajo at my old shop was geared in exactly the way you just described 6 months after I left. Incidentally, it was a very experienced pilot, he just got distracted and focused on the wrong things at the wrong time. He's a good guy - but even the guy who geared up the airplane says, "yeah...it was pilot error."

Similarly, there are lots of Pilatus pilots out there who love to land with the flaps at 15°. This makes for some greaser landings, but it's really easy to gear up the airplane with the flaps at 15° because you can silence the warning horn and coast in. In my opinion, you're better off making a flaps 30° or 40° your standard operating procedure, because when you add your last flaps in at 500' AGL you'll have a lot longer to thinking about "what the heck that annoying noise is."

 

[milleR]

What will be particularly interesting is whether the corporate manufacturers attempt to get single pilot certification for new airframes based on the capabilities of the software/avionics. If we see a G750 Single Pilot type rating become an option then the writing's on the wall.

 

[MikeD]

@MikeD

A Navajo at my old shop was geared in exactly the way you just described 6 months after I left. Incidentally, it was a very experienced pilot, he just got distracted and focused on the wrong things at the wrong time. He's a good guy - but even the guy who geared up the airplane says, "yeah...it was pilot error."

Similarly, there are lots of Pilatus pilots out there who love to land with the flaps at 15°. This makes for some greaser landings, but it's really easy to gear up the airplane with the flaps at 15° because you can silence the warning horn and coast in. In my opinion, you're better off making a flaps 30° or 40° your standard operating procedure, because when you add your last flaps in at 500' AGL you'll have a lot longer to thinking about "what the heck that annoying noise is."

Agreed. And on the Navajo, I was just pulling out a random example of being distracted on landing for a random (important) reason, and missing the gear. It happens, as you saw. And like you say, it's not that the guy was a bad guy or bad pilot in any way. He made a mistake and unknowingly allowed a distraction to interfere with his standard game plan. It could be that there was some unusual reason for why this happened, but often, there's no "magic bullet" of why the distraction was allowed to interfere with someone's habit pattern to the degree that it was a contributory factor to a mistake by the pilot that ultimately was causal to the accident. Sometimes these things happen.

 

[ppragman]

Agreed. And on the Navajo, I was just pulling out a random example of being distracted on landing for a random (important) reason, and missing the gear. It happens, as you saw. And like you say, it's not that the guy was a bad guy or bad pilot in any way. He made a mistake and unknowingly allowed a distraction to interfere with his standard game plan. It could be that there was some unusual reason for why this happened, but often, there's no "magic bullet" of why the distraction was allowed to interfere with someone's habit pattern to the degree that it was a contributory factor to a mistake by the pilot that ultimately was causal to the accident. Sometimes these things happen.

This. A million times this.

In my opinion, two of the biggest tools a PIC can have in his or her toolbox is self-awareness and "big picture" thinking. Awareness of what your focused on gives you the ability to say, "Am I paying attention to the right things?" One of the things I see guys do pretty commonly is start pushing buttons on the FD A/P when they should be flying the airplane, and flying the airplane when they should have George on. Not that there are really any standards as to when one is better than the other, but hand-flying when you need to be running checklists is just as silly as trying to reprogram the autopilot when it fails to level off instead of simply clicking it off and manually flying the level-off. Having an understanding of where you're focusing your attention means you can choose where it's being focused, instead of simply just having it directed by the pretty lights and aural annunciations. Similarly, "big picture" thinking (and I don't just mean situational awareness) means that innate sense most guys have of when a particular battle is no longer one you want to fight anymore. When do you go around? Most people say, "when stabilization criteria are exceeded" - and yet, aircrews all over the world have ignored those "limits" and proceeded to ball up airplanes. Big picture thinking means that you have to be ready to give up a particular battle and focus on winning the "war" (that is to say, not crashing). Maybe I'm talking out of my ass - God knows I'm a bit of a hack, but these are observations that I've had over the years of doing this job.

 

[PositionAndHold]

What will be particularly interesting is whether the corporate manufacturers attempt to get single pilot certification for new airframes based on the capabilities of the software/avionics. If we see a G750 Single Pilot type rating become an option then the writing's on the wall.

If I'm a guy with the means to have a 65-70 million dollar jet, there's going to be two hearts beating up front. If one quits beating, we've got another. That's what I've never understood about owners of single pilot citations etc. They have have all the redundancy, two engines, dual electrics/hydraulics but we're just going to have one pilot. Whhhaaat?!? Pilot incapacitation is rare but it can happen.

As a side note, how much of the avionics in UAS systems meet part 25 or even part 23 standards? There's a reason you can get a cheap autopilot/synthetic vision for your RV and not for your 172. Certification standards are different and they cost money. Sure, UAS systems have the capability to fly a large passenger jet remotely but can they do it to part 25 standards? That's not just the boxes but the code in the firmware/software. I'm guessing that will mean at least three separate data links and all the programming a certified avionics system goes through. There's no way that will be close to cutting the cost of what it takes to put a second body upfront. Not any time soon. We can talk about cargo doing it first sure, but does any one think the cargo cut out would exist today if UPS took out that neighborhood a few hundred yards short of the last crash in BHM? Or if UPS 6 hit a a neighborhood instead of the only dirt lot in the area in Dubai? We've been lucky not good.

 

[Vector4Food]

Interesting thread.

I would like to point out, that as usual, the understanding of the ATC system is rudimentary at best in this thread.

My field has been implementing automation for a long time, with the purpose of reducing staffing.

Guess what? About two years ago some systems that were intended to remove bodies, actually saved zero staffing, but dramatically increased the safety factor.

Without a quantum leap in technology, humans will still be the forefront of aviation for the foreseeable future, in conjunction with good technology.

Thinking otherwise is just a pipe dream

 

[seagull]

No, but you can certainly train people better so that they're more able to solve problems in the airplane in a manner that maximize the possibility of success.

We can do a better job if we give the pilots the tools to do it. Unfortunately the trend is to reduce training and that is partly based on the idea that we have designed systems that will protect us so the pilot does not need those skill sets, and a large part of that is that part of the time we can, it just fails when there are additional factors going on such as stress, high workload, fatigue, etc. It is just that usually those factors do not combine. So, yes, we need to give the pilots the skills so they can provide the resilience.

 

[seagull]

If he crashed almost crashed on a visual approach...then yes, he is weak at flying visual approaches, he should receive more training in them.

The problem was not the visual per se but the combination of an unexpected system function (the autothrottles having a corner point that they would be on but not respond that was not known by Boeing either and not well advertised after it was discovered, and not trained in any event), when coupled with the other factors.

 

[wheelsup]

The Asiana approach was fine. He was right on path until 1-2 miles out, when he started drifting below GS. It wasn't like he was all over the place for the entire approach.

From my understanding it wasn't a lack of hand flying ability but in fact the opposite, a weakness in automation proficiency on that aircraft.

 

[seagull]

In the sense that ultimately, it's the job of the pilot to safely execute the flight, its their error. A secondary factor would indeed be "might we have a safety/warning system that isn't as effective as we believed it to be?"

If a pilot fails to lower the landing gear on an approach to a runway in high crosswinds, and he's so engrossed in mentally concentrating on the landing that he lands gear up, and only when the sliding stops does he notice a gear warning horn blaring at him, then the primary causal factor is indeed the failure of the pilot to lower the landing gear (and the related reasons of how/why that occurred), a secondary factor would be the warning system and did it work as advertised? Is it truly effective? Etc.

Because ultimately, it is the pilots responsibility to lower the landing gear for landing. Bottom line of PIC responsibility.



I'd look at training. Your descripton of the above regards the autothrottles, reads to me that this is a known potential problem/issue with guys who transition from aircraft type X to aircraft type Y, or manufacturer type X to manufacturer type Y. If that is the case, then did the training program the crew received cover this extensively as a known "gotcha" that has bitten crews past? How much was this emphasized in ground, simulator, an/or course study materials?

The error isn't necessarily in the system, IF the pilot is either unfamiliar with the system, doesn't know its nuances, or doesn't utilize the system as its supposed to be used.

Ultimately, the crew made the error and crashed a perfectly good operating airplane. The why of it all, could reveal other issues such as the above.

Asiana case it was just a small note in the manual and only added just prior to the accident. Most pilots flying the Boeing aircraft with this architecture were not aware of it when surveyed after the Asiana accident, and that tells you something.
With regard to your statements on pilot error, that is much how I used to believe also. I highly suggest you start by reading Dekker's most recent edition of "A field guide to human 'error' and then write me your thoughts after you have read it. I would be interested in your opinion.

 

[seagull]

That's exactly what he's saying.

No, it is not. Humans have a partiality towards simple explanations which leads to the way we think about causality We have become so accustomed to it that we no longer even notice that we're doing it. Things look much simpler in hindsight because we have a clear view of how a decision led to something else, but if we were to be able to look at all the factors the person was actually experiencing at the time we would see that the actions are almost always not so clear cut and the consequence of the action is unknown at the time.