Airbus A320 crashes in French Alps

[Subyman]

Frankly it is so rare, at least as far as I know, that we really can't draw any conclusions. All I know is that if I'm on a plane with a pilot that has an issue with depression I would sure as hell rather him being treated for it than hiding it. Granted in a perfect world I would choose neither scenario but a perfect world it is not.

Have there been other commercial airline crashes due to the pilot being suicidal (excluding terrorism/hijacking)? Most people that commit suicide just kill themselves. A small percentage take a specific person or persons with them but aside from suicide bombers the vast majority of suicidal people don't kill innocent people they don't even know.

At the end of the day we probably need some sort of compromise. Like quarterly psych evaluations for any flight crew with history, or displaying signs, of depression.

I don't think there is much we can do either. Problem will fix itself when autopilot does the entire route. I think it is tragic when millions of man hours are spent engineering redundancy in every system of the plane, developing training techniques to best problem solve in-flight situations, and coming up on a century of flight innovation only to have that nullified by the desire of one depressed man to direct the plane directly into the ground.

No amount of training or engineering will work around the desire of a man to kill himself, until we take the human out of control.

 

[crashtech]

I wonder at that point how the statistics will compare between human and computer. Software and hardware are not without their vulnerabilities, obviously. Right now, humans have a fairly decent record, and are able to accurately parse situations that are beyond most programmer's imaginings.

 

[DCal430]

According to leaked information he may have been prescribed Zyprexa and Ativan, neither of which are SSRI. One is an anti-psychotic and the other is an benzo. Scary if this is true.

 

[Subyman]

I wonder at that point how the statistics will compare between human and computer. Software and hardware are not without their vulnerabilities, obviously. Right now, humans have a fairly decent record, and are able to accurately parse situations that are beyond most programmer's imaginings.

Computers are only as good as their input as well. Hard to simulate the human feel of the airplane. A lot of times in dire situations its the pilot's thousands of flight hours seat-of-the-pants know-how that saves the flight. Something as dumb as the autopilot using a faulty pitot tube can crash an airplane, such as flight 301.

 

[Caravaggio]

I wonder at that point how the statistics will compare between human and computer. Software and hardware are not without their vulnerabilities, obviously. Right now, humans have a fairly decent record, and are able to accurately parse situations that are beyond most programmer's imaginings.

Inclined to agree with you.
Let's imagine some scenarios where a computer might struggle.

1) crosswind landings required in very gusty, low visibility, conditions when the microwave is down and you cannot get fully established.
2) responding to ATC emergencies and communicating with other aircraft.
3) knowing when NOT to transmit across another pilot's emergency call.
4) quick response to power failure at V2 on rotation.
5) dialling in 7500 on the transponder. Would the computer even know it had been hijacked?
6) emergency go-arounds when ATC must be instantly trusted without time for verification.
7) responding to an all-engines bird-strike at full power.
8) helping another pilot with a 7600, to relay information back to ATC.
9) making a judgment about a critical sea state in a DHC Beaver floatplane, approaching Ganges BC, in bad weather, as the sun goes down.
10) Giving way to a sudden 7777 squawk in your flight path.

I would trust the human above the computer in most of these situations.

 

[Darwin333]

I don't think there is much we can do either. Problem will fix itself when autopilot does the entire route. I think it is tragic when millions of man hours are spent engineering redundancy in every system of the plane, developing training techniques to best problem solve in-flight situations, and coming up on a century of flight innovation only to have that nullified by the desire of one depressed man to direct the plane directly into the ground.

No amount of training or engineering will work around the desire of a man to kill himself, until we take the human out of control.

Oh we have that capability already, I just don't think the public is ready to fly in a plane that is completely computer operated. Most modern jumbo jets actually have a "button" that you can push and the autopilot will land the plane all by itself.

Almost as big of a problem as public acceptance is the FAA. The airlines will have to spend an ungodly amount of money just on FAA testing alone and that doesn't include costs of integration and their own testing. We still haven't integrated technology into the planes that lets us know exactly where it is at when an emergency occurs leading to a crash. I can personally build a system that would do exactly that for cheap as hell. It doesn't need to survive the crash or anything just transmit it's current location when an alarm (like the "Warning-terrain-pull up" on the current crash). We still haven't integrated what has now become very common and cheap technology that we all carry around in our phones. Planes can't send and receive real time information to and from each other, my phone can tell me where my friends are in real time and vice versa. It's crazy that we can lose a jumbo jet for months.

 

[Subyman]

Oh we have that capability already, I just don't think the public is ready to fly in a plane that is completely computer operated. Most modern jumbo jets actually have a "button" that you can push and the autopilot will land the plane all by itself.

Almost as big of a problem as public acceptance is the FAA. The airlines will have to spend an ungodly amount of money just on FAA testing alone and that doesn't include costs of integration and their own testing. We still haven't integrated technology into the planes that lets us know exactly where it is at when an emergency occurs leading to a crash. I can personally build a system that would do exactly that for cheap as hell. It doesn't need to survive the crash or anything just transmit it's current location when an alarm (like the "Warning-terrain-pull up" on the current crash). We still haven't integrated what has now become very common and cheap technology that we all carry around in our phones. Planes can't send and receive real time information to and from each other, my phone can tell me where my friends are in real time and vice versa. It's crazy that we can lose a jumbo jet for months.

Autopilot still buggers off if something outside of its operating parameters comes up, which is why the pilot still needs to be there. Shouldn't be much longer though. I give it about 20 years before we see some type of automated commercial flights. I'd think cargo planes would be the first to switch over.

 

[crashtech]

Autopilot still buggers off if something outside of its operating parameters comes up, which is why the pilot still needs to be there. Shouldn't be much longer though. I give it about 20 years before we see some type of automated commercial flights. I'd think cargo planes would be the first to switch over.

There's been some accidents caused by autopilot commanding to the limits of its authority and then suddenly handing a worsening problem to the human crew without them being at all aware that there was a serious problem with the aircraft.

If we humans weren't so squeamish about certain things, I bet we could achieve a much greater synergy between our brains and automated systems that could serve to improve safety and have many other benefits, including optimized fuel consumption and the smoothest possible ride.

Honestly I am skeptical about AI going it alone in the near future. Drone programs use operators on the ground, it's not inconceivable that we could see other aircraft utilizing something similar, especially when the flight unintentionally exceeds the operational capabilities of the autopilot.

 

[Subyman]

There's been some accidents caused by autopilot commanding to the limits of its authority and then suddenly handing a worsening problem to the human crew without them being at all aware that there was a serious problem with the aircraft.

If we humans weren't so squeamish about certain things, I bet we could achieve a much greater synergy between our brains and automated systems that could serve to improve safety and have many other benefits, including optimized fuel consumption and the smoothest possible ride.

Honestly I am skeptical about AI going it alone in the near future. Drone programs use operators on the ground, it's not inconceivable that we could see other aircraft utilizing something similar, especially when the flight unintentionally exceeds the operational capabilities of the autopilot.

I'm wondering though if you take the pilot out of an aircraft holding 200+ people then they don't have as much skin in the game. Drones are great for the military, you don't lose a pilot if they aren't in the plane when it goes down. But with public transportation, you'd think having the pilot on board would give them the extra tactile feel of the aircraft that would help in a crisis situation.