Which would be more dangerous?

[Shiny Side Up]

Completely autonomous flight is possible to do now. So what?

I'm just glad you acknowledge that point, it means we can get somewhere with this arguement. You seemed to indicate you believed otherwise before, so we're making progress.

Lots of things are technically possible if you're willing to ignore the risks.

And human beings have had a long history of damning the risks in the name of progress. The amount of people who've died in the name of making aviation what it is today is long. Do you believe that we're done with the blood sacrifice of progress? Icarus might have been the first, but Michael Alsbury won't be the last.

[Rockie]

You seemed to indicate you believed otherwise before, so we're making progress.

I've never implied it wasn't possible now so I don't know where you get that idea. It was several years ago now that a large drone flew autonomously from California to an air force base in Australia.

And human beings have had a long history of damning the risks in the name of progress. The amount of people who've died in the name of making aviation what it is today is long. Do you believe that we're done with the blood sacrifice of progress? Icarus might have been the first, but Michael Alsbury won't be the last.

Explorers and adventurers take those risks, not the traveling public. For instance you could have ridden inside the drone that flew itself to Australia and you would have been fine, but do you think TC would approve an OC to carry passengers in it? Why not?

[complexintentions]

I'm still waiting for the flying cars in every household we were promised in the 1950's. Also completely within the realm of the possible technically, yet...?

Whimsical aphorisms like Dylan's "The times they are a-changin' " aren't really substantial enough to defend autonomous airliners. Sorry.

What's ironic is that I WISH that fully autonomous flight were possible (in the sense of technically, economically, politically, psychologically viable, today, in 2015, not some vague time in the future) - yet some of us are forced to try and explain to the self-styled "futurists" how it's not quite as close as they assume.

-People who have done no market research, stating we "must admit there is probably market demand for such a thing right?".

-Pointing to self-parking cars, and airport trams, and the Space Shuttle missions and drawing direct comparisons to the commercial passenger flight infrastructure that move millions every day.

-Using fly-by-wire as an example of our trust in computers. When it's nothing of the sort - FBW is only there to act as an interface between the human operator, manipulating the machine either directly or through the selections of the autopilot. It does not possess any "intelligence" per se. We depend on the computers it uses to correctly send signals based on what it is directed to do - it does not make decisions in any meaningful way and we certainly don't "depend" on it to act autonomously or without error or failure.

-Equating technology whose failure causes inconvenience, with technology whose failure costs lives, and giving both equal weight.

And on and on with arguments that not only don't make logical sense, but can't really be refuted because they're based more on gut feelings formed from randomly collected perceptions. "Predator Drones! Mars Missions! OnStar! iPhones! We MUST be close to pilotless airliners! In 20 years! And if you ask me how I got 20 years, not 12, or 23, I'll just pull another answer out of my ass!"

Nat Geo recently had an article called "The War on Science" that articulates this growing phenomenon of basing one's beliefs on feelings rather than rational logic. It's kinda like we're heading into an "anti-Enlightenment" period, where either we reject science completely (anti-vaxxers, climate change deniers) or embrace it unquestioningly without acknowledging its limits. (*ahem*) Since I don't believe in endlessly fighting battles that aren't winnable, I'll bow out and let Rockie keep slugging away.

But I'll leave this thread with one last thought. It isn't about designing a computer or system of computers that's "reliable" enough. The failure rate of modern airliners is already astoundingly low, well below the threshold required to gain public trust. That isn't the issue. But even the most sophisticated aircraft flying are still primitive in the sense that they lack the ability for critical thought.

Currently, automated systems on commercial aircraft can't "think". Barring mechanical failure, they do exactly what they are programmed to do. I always laugh when people cry out for more automation after an accident that is blamed on "pilot error". Inevitably the accident was caused by the pilot ceasing to think critically: i.e. why did the Asiana pilot not question why it was requiring so much control back pressure to hold altitude on the approach to SFO? (As the airspeed decayed, with disastrous results). In other words, the accident could have been avoided if he acted LESS like a machine - mindlessly holding altitude, and more like a human and exercised the ability for critical suspicion.

So far, that ability, at its most sophisticated, is uniquely human. Closely mimicking it with brute processing power may work for non life-critical tasks, but it doesn't cut it for autonomous flight. If that changes one day in the future, the implications of having cognizant, self-aware machines are far more far more far-reaching than just aviation.

[Blueontop]

Mic drop!

[Shiny Side Up]

I'm still waiting for the flying cars in every household we were promised in the 1950's. Also completely within the realm of the possible technically, yet...?

Whimsical aphorisms like Dylan's "The times they are a-changin' " aren't really substantial enough to defend autonomous airliners. Sorry.

It wasn't supposed to. It was merely to say it would be unreasonable to count out such possibility. Flying cars are actually a longer possibility, that's never really had any feasibility, but you never know. It would be foolish to bet that there would never be flying cars, at least if one was going to extend that bet beyond your lifetime. Oddly enough, for enough naysayers, they keep trying to make them.

What's ironic is that I WISH that fully autonomous flight were possible (in the sense of technically, economically, politically, psychologically viable, today, in 2015, not some vague time in the future) - yet some of us are forced to try and explain to the self-styled "futurists" how it's not quite as close as they assume.

I would only argue that it may not be as far as you assume either. I personally don't think I'll see it in my lifetime, but hey, I also don't know how long I'm going to live, that statistically keeps getting longer for the average earthman. One of the points of the arguement though is that you don't need to have fully autonomous flight to have pilot-less flight.

-People who have done no market research, stating we "must admit there is probably market demand for such a thing right?".

Actually in reference to an OnStar type safety system which might extend into being able to somewhat control the aircraft, I'm not that far off. Given that (since I was counting) I've managed to train somewhere in the hundreds of pilots, such safety systems are desirable in a good majority of the private flying crowd, where safety is a premium. Press-button-copilot is a desirable item, just like BRS and airbags, triple redundant GPS and autopilots.

-Pointing to self-parking cars, and airport trams, and the Space Shuttle missions and drawing direct comparisons to the commercial passenger flight infrastructure that move millions every day.

Actually no, not direct comparisons, current applications that are comparable to the hurdles proposed against. So far I've been told that the general public is well educated and paranoid of computers. Not sure, but wander around the concourse of any airport and I'm sure you'll have a different opinion. You guys do know that most of your passengers already think a computer does most of the work, right?

-Using fly-by-wire as an example of our trust in computers. When it's nothing of the sort - FBW is only there to act as an interface between the human operator, manipulating the machine either directly or through the selections of the autopilot. It does not possess any "intelligence" per se. We depend on the computers it uses to correctly send signals based on what it is directed to do - it does not make decisions in any meaningful way and we certainly don't "depend" on it to act autonomously or without error or failure.

Again the idea that it needs to have some sort of advanced AI to function, or that there won't still be a human input. You're right, real AI is technology a long ways off. You just don't need it to fly an airplane from point A to B. I hate to tell you this, but if you talked to a lot of pilots, you'd find there's a disconnect with what a pilot's job is. I've run into enough people who don't think it has anything to do with stick and rudder anymore, and the appearance of the new MPL is somewhat worrisome. Oddly enough, automation-wise, taxiing the thing is the most difficult hurdle to overcome, paradoxically the easiest job of the pilot (or at least you'd hope, I've seen enough airplanes taxied into hangars, fences, over runway lights, into other airplanes....)

growing phenomenon of basing one's beliefs on feelings rather than rational logic.

Well, I guess both sides of this argument feel that way.

[Rockie]

What kind of flying do you do SSU?

[Shiny Side Up]

What kind of flying do you do SSU?

Ah, good true to form, whenever we have these fun talks, time to attack my credibility.

To your great satisfaction, I am not an airline pilot. How can I possibly know, right? To answer though, and give you food for thought, I spend a fair amount of my time when flying airplanes, doing survey work. In the past it was instructing, which still is still a majority of my time in airplanes. In relevance to this discussion, my work as a survey pilot is definitely days numbered when it comes to competition from UAVs. I can accept that. While I would like to think that there's a lot of areas where my expertise still trumps that of the machine, they don't trump it that much, and I don't see that they will for long. For example, as a person I can be somewhat more efficient with the plane, end turns on lines I can do in less time and space, than Otto can, but lets be clear, Otto definitely has the capability to catch up. As a small example. Much work on some fronts already has stiff competition from drones and all sorts of other related and similar devices. The only issues are currently regulatory - to which there is pressure to solve. Big bucks are involved. I'm keenly aware that in a lot of cases I still have a job because I'm still cheaper, but I don't expect that to last. At the very least they'll need to send me along to make sure the thing can be fueled.

To note one of the sectors of flying you can expect to disappear is doing oblique work, since it can be easily done by UAVs. Not rocket science.

Now I know, you'll jump on how your flying is different than mine, but how much is it really? You have passengers is at its core the only difference. Otherwise, you're still flying a machine from A-B (actually my flights are more like A-b1-b2-b3-b4-b5-... well you get the point) we both have all the difficulties with weather, traffic, mechanical issues, etc.

Oddly enough, I would foresee in the long future, that flight instructor might be the only pilot job around. Hopefully there will still be people interested in making homebuilts, and booting around the patch, probably restricted to the hideously wealthy. And probably restricted heavily to avoid the UAV traffic.

Now I suspect the only difference between you and I is that I know there are interested parties looking into eliminating my pilot job. IF you think the question to eliminate yours isn't being asked by those with money to make it happen, well you're just kidding yourself. They've already made it legal to get rid of two pilots. I don't think it will be long until you might have several people on board who maybe between them all are one pilot. Or maybe the only one who can actually land an airplane is some guy with a PPL in a passenger seat.

[CID]

For the most part I've said my piece and I don't think there is any sense trying to discuss this with the more emotional and seemingly less technological side but I want to make a statement against a remark about FBW.

FBW doesn't simply translate the input commands from the pilot and apply them to the control surfaces. That only applies after a failure and the system reverts to direct law. in normal law and to a lesser extent in the alternate law (a degraded mode) it inserts all manner of protection to prevent the pilot from doing the wrong thing. The system monitors flight parameters and aircraft configuration. It's a much more automated process than simple control translation.

And how about FADEC? FADEC engines have basically made starting and operating turbine engines as simple ans flipping a switch and setting the power. No more multi-step starting procedures or manually setting trims and such. In fact, having a human tweaking an engine like a FADEC computer would require another person in the cockpit but the flight engineer has gone the way of the Dodo. One more down, two to go.

[ragbagflyer]

Rockie, the open letter you linked to wasn't actually the letter, but an attached "research priorities document". That document was drafted by a few, with the input of several more, and attached to the open letter which has hundreds of signatories. None of which are anti-AI as far as I can tell BTW. They're simply acknowledging a need for consideration of the issues that will arise as we march closer to AI.

This thread could really be split into two questions. Will airliners be pilotless prior to full-blown AI and will airliners be pilotless after full-blown AI? As far as I'm concerned to only question worth even debating is the first question, because once we've hit full AI we are objectively redundant by every measure.

Even now there's really not much we can do better than a machine. We make slower calculations, we're greatly susceptible to 'human factors' (and there's a TONNE of human factors), procedurally we're weak, our memories are total shite; and as much as it pains me to say it we don't even have better hands and feet anymore. Just look at the built in stabilization on your typical $150 remote control aircraft.

The only edge we have left is the ability to reason and once we lose that we're toast. Let's get hypothetical. Indulge me for a minute and agree that AI is achieved and finds its way into a pilotless airliner. Let's say it's a Boeing. Everyday each of those 'brains' is going to accumulate a human lifetime of experience by taking in the experience of every single other Boeing flight that day. Every emergency experienced by every singe Boeing that day can be 'experienced' by all of them. We think of our own experiences in the cockpit as valuable (and they are) because we've spent our entire careers accumulating them and as of now there isn't a method to share an experience; but once there is the whole concept of what an experience is changes completely.

Eventually the public will have no objection because they'll have already seen almost every other profession be superseded by a more capable machine. Surgeons, engineers, drivers, mechanics. Nobody is immune.

Perhaps we never achieve all this sci-fi craziness. I don't think I'd make that bet.

None of us should be so arrogant as to imagine that we can fathom every possible advance and/or catastrophe that the next 50 years, or even the next 25 years is going to bring.

[Rockie]

Ah, good true to form, whenever we have these fun talks, time to attack my credibility.

Actually the reason I asked what you flew is so that I could try and frame the discussion around your experience instead of around mine. Try and not be so sensitive.

You see, no matter what a pilot flies he/she has to use situational awareness, subjective analysis, predictive reasoning and risk assessment in making decisions. "Judgement" is a subjective thing derived from many sources that humans have and computers do not, including intuition and emotion. I submit that you use all of these things on a continuous basis when you provide training to another pilot. You make decisions based on a lot of factors you probably aren't even consciously aware of when you're doing your survey work. The actual mechanics of controlling the airplane is only one part of it, and I further submit that logarithms can never be as good as you are doing even that because they have no finesse. They cannot predict something that hasn't actually happened yet and modify their behavior ahead of time. Computers make a good tool, but the world's worst pilots.

Given that I have over 30 years using the kind of technology we're talking about can you not allow that I might - just might - know a little about it?

Rockie, the open letter you linked to wasn't actually the letter, but an attached "research priorities document". That document was drafted by a few, with the input of several more, and attached to the open letter which has hundreds of signatories. None of which are anti-AI as far as I can tell BTW. They're simply acknowledging a need for consideration of the issues that will arise as we march closer to AI.

It was signed by many including the three I mentioned plus more equally as intelligent and intimately acquainted with the issue, then endorsed by hundreds more people with lots of relevant letters after their name. No, they aren't anti AI, but they are adamant that it should always be subordinate to human judgement when it comes to critical decision making. They use vehicles (simple ones like cars - not airplanes) and weapon systems as two cases where human superiority must be maintained for many reasons.

[Rockie]

Even now there's really not much we can do better than a machine.

Completely disagree. Machines are tools...nothing more. They wouldn't even exist if humans didn't create them for a clearly defined, narrow purpose.

Let's get hypothetical. Indulge me for a minute and agree that AI is achieved and finds its way into a pilotless airliner. Let's say it's a Boeing. Everyday each of those 'brains' is going to accumulate a human lifetime of experience by taking in the experience of every single other Boeing flight that day. Every emergency experienced by every singe Boeing that day can be 'experienced' by all of them.

I've asked this question before, but who says this AI won't make mistakes and kill lots of people on a regular basis?

[Captain S itmagnet]

All hypothetical of course... On a day in the future...

"Captain" Woo, with all of 150 hours in a "real" airplane is in the boiler room/pilot centre operating a flight on behalf of Really Modern Airlines with shite weather at destination and an MEL item that might be a bit grey as to whether he should continue or not. Let's say hmm how about some deice/anti ice failure. Throw in what had he really been there what his real eyes would see - a line of crap building ahead that's not yet on weather radar.

Now the airline's rep from "Commercial" is looking over his shoulder with some not so subtle pressure to continue. Strange, because yesterday no one from Very Modern Airways watched his decision making and Extremely Modern Airlines reps have never been seen in the room.

Anyone care to finish the story?

[Shiny Side Up]

You see, no matter what a pilot flies he/she has to use situational awareness, subjective analysis, predictive reasoning and risk assessment in making decisions. "Judgement" is a subjective thing derived from many sources that humans have and computers do not, including intuition and emotion. I submit that you use all of these things on a continuous basis when you provide training to another pilot. You make decisions based on a lot of factors you probably aren't even consciously aware of when you're doing your survey work.

While training people is one thing that does constantly require my judgement and assessment, flying point to point rarely does. Most decisions are made on the ground before one goes, and only on occasion does the flight need my decision powers, and rarely are those critical to the safety of flight. Try not to over-glorify what we do. Realistically, the poor soul who rides in the back makes most of the operational decisions, and the main things I'm called upon to do is get it off the ground and back. Life's easy when you don't have to interact with human beings, the amount of unforseen things really shrinks. Operationally, many of the things not to need pilots would involve changes in procedure, which of course might be easier if there weren't emotional human beings to deal with.

I've asked this question before, but who says this AI won't make mistakes and kill lots of people on a regular basis?

Why does such a thing need AI? You know that the US military is training more drone pilots than they are other pilots, right? Predators and Reapers aren't run by SkyNet. Also one might note that they're sharing the airspace now. Here in North America. How do they do this? Why aren't they smashing into each other or icing up and falling out of the sky? Does Northrop Grumman and General Atomics have some sort of secret AI program the public doesn't know about? Or are the solutions much simpler than that?

[Rockie]

While training people is one thing that does constantly require my judgement and assessment, flying point to point rarely does. Most decisions are made on the ground before one goes, and only on occasion does the flight need my decision powers, and rarely are those critical to the safety of flight.

Well then, I can only say that your experience is extremely limited. Once again can you not allow that some people who actually do operate in roles and environments beyond your experience may know more about this than you do?

BTW never heard of General Atomics, who are they again?

[photofly]

It's interesting that great leaps in progress sometimes come from people who don't have experience in the field. Everyone who "knows how difficult it is" is too busy saying how difficult it is (and protecting their patch) leaving the field wide open for someone who can see how to do it.

[CD]

BTW never heard of General Atomics, who are they again?

General Atomics Aeronautical Systems, Inc. (GA-ASI)

The business of General Atomics Aeronautical Systems, Inc. (GA-ASI) is the development of transformational technologies that deliver paradigm-changing results. An affiliate of privately-held General Atomics, GA-ASI is a world leader in proven, reliable Unmanned Aircraft Systems (UAS) and tactical reconnaissance radars, as well as advanced high-resolution surveillance systems. The company is dedicated to providing long-endurance, mission-capable aircraft with the integrated sensor and data link systems required to deliver persistent situational awareness and rapid strike capabilities.

The company's Aircraft Systems business unit designs and manufacturers game-changing, life-saving unmanned aircraft systems, including the ground stations that control them and provides pilot training and support services for field operations. Revolutionizing aviation by expanding the capabilities of UAS, the business unit develops innovative high-tech solutions that have produced an ever-growing line of versatile, reliable, cost-effective, and combat-proven aircraft.

[Rockie]

Try not to over-glorify what we do. Realistically, the poor soul who rides in the back makes most of the operational decisions,

Why are passengers on your airplane making operational decisions? They can make operational requests but the decision whether or not to do it rests with you. That isn't over-glorifying the job - it is your responsibility as PIC by edict of the Parliament of Canada. You couldn't give that responsibility away even if you wanted to nor would it be smart.

[Shiny Side Up]

BTW never heard of General Atomics, who are they again?

I rest my case.

[DonutHole]

Rah rah siskoombah
http://dronewars.net/drone-crash-database/

What could go wrong.

A quick survey of the issues leads me to believe remotely operated aircraft do nothing but add complexity and modes of failure. You still have pilot error , mechanic failure, weather and you add the 'lost link' scenarios...

There's hundreds of crashes over a five year period...

Not gonna happen packing hundreds of civilians across the .? I just don't believe as humans we make good enough machines to do this safely on the scale of large airliners and the associated numbers of movements.

But general atomic says they're really really good at this stuff... and I never question a sales pitch.

[Rockie]

BTW never heard of General Atomics, who are they again?

I rest my case.

Maybe you're right. Letting a computer make the operational decisions in your plane might be a better idea than your current practise of getting your passengers do it.