rickair7777

That's part of the economic problem with automation... no grey area.

An automated airliner will divert or cancel, at great expense, for trivial things like the one of the lav light bulbs burned out or the GS fluctuates momentarily in CAVU.

PlaneS

This technology, in and of itself, is not too impressive.

We have gone (*knock on wood*) over a decade without a fatal accident in the US by a US carrier (SWA 1380 notwithstanding, but that was nothing the pilots could've controlled).

The machine learning just isn't anywhere close to where it has to be to replicate that safety record. For at least the next generation or two, I can't see a good replacement for an experienced, well-trained crew.

Critics will point to the fact that the vast majority of accidents are caused by pilot error, and that's true, but robots are nowhere close to being as skilled as pilots, nor do I think they will be in our lifetimes. Replacing human crews with unsatisfactory robots will result in either 1) catastrophic accidents, or 2) unnecessary loss of revenue as situations that could be mitigated by human crews are just handled by blanket return-to-gate/diversion/precautionary landing solutions.

The unique scenarios the AI will have to handle are nearly infinite. What if an MD-11 is crossing at the threshold and the glideslope starts acting up, showing up to 1 dot above or below - will the computer know this is an error, or will it think it's unstable and go-around? What if you are on final and see a microburst hit the airport with no warning from ATC or your onboard equipment (something that happened to me) - will the computer continue? What about the 2005 BOS runway incursion - how would the AI decide to gun it and try to lift off at Vmu, or hold the nose down until way past Vr and rotate with only a few hundred feet left? Or the Hudson River landing? I shudder to think what a computer would have done during the 1999 PVD runway incursion, when ATC gave the takeoff clearance but the US Airways crew decided to hold their takeoff and saved the lives of everyone on two planes.

The point I guess I'm trying to make is that you can't program a computer to respond to every one of these situations individually. It has to be able to learn, observe, and make split-second decisions on its own. And I just don't think the technology is anywhere near where it has to be to handle all these situations safely. I wouldn't be worried, even for the cargo pilots; you are sharing the same airspace as the rest of us, above the same densely-populated cities. If one of us is going to go automated, it means we all go.

It's not worth getting worried about (yet).

ObadiahDogberry

So within days after I posted this, I have seen two examples of AI/algorithms making mistakes. First off, a cabin crew friend of mine who has a few thousand followers on her Instagram account just got her account suspended because the Instagram algorithms have determined she is a "fake". I know this woman quite well (although as much as I might have liked, I haven't known her in the biblical sense), have flown many trips with her, and we regularly chat. She is 100% real, but AI has determined that she isn't. Same thing happened with an old high school class mate of mine who is a professional singer and recording artist (he is not super famous, but does well enough to make a decent living). For some reason Facebook determined he was impersonating himself, and shut his professional account. He has used his professional Facebook account for seven or eight years now to promote his albums and his tours. His messages to Facebook support have gone unanswered beyond some automated responses. Understandably he is upset about this development because it does affect his livelihood, especially now that he cannot tour due to Covid 19. So just two pieces of anecdotal evidence of AI from multi billion dollar global tech companies screwing up some relatively simple things, and they aren't even life or death decisions.

HotDogSonicBoom

What do y’all see here?

https://youtu.be/piYnd_wYlT8

We are a long way from losing our seats. Never mind the p=np or p!=np debate... that’s a friggin turtle in that video.

Bigapplepilot

I do think he was talking about ‘Evtol’, ‘flying taxis’, or whatever they’re called today. That system would probably be fully autonomous with a completely separate autonomous ATC system operating drones in a fixed system (very limited) for Urban transport(400 feet and below). Not a Part 121 airliner flying NY to London.

rickair7777

Those can be autonomous. If something goes wrong, they just land. Or maybe crash, but they won't kill more than a tiny handful of people. With no judgement calls to make, that's a pretty easy nut to crack.

Like they can automate the train that runs on closed tracks between terminals at the airport, but they can't automate long-haul pax or freight trains that operate in the real world.

takingmessages

Oh, dear...

https://www.reuters.com/video/watch/...?chan=dxefo27m

AirBear

Not gonna happen, I mean who would write the liability insurance for something like that? Even Owner Operated Turbine equipment is getting difficult and very expensive to find coverage for. First time something happened that the computer couldn't handle and that cargo plane winds up in someone's living room it's game over.

rickair7777

Here's what Fred Smith has to say about large autonomous airliners...

"This initiative deals with small turboprop airplanes and in this particular case the single-engine C208, which we’re looking at putting in very remote and uninhabited areas as part of our network,” Smith said Monday at FedEx’s annual shareholder meeting.

FedEx pilots shouldn’t be concerned about robots stealing their jobs -- for now. It would take decades for technology to replace humans in piloting large freighters, Smith said.

"FedEx has no plans to replace its trunk aircraft fleet with autonomous aircraft,” said Smith, a former pilot in the U.S. Marine Corps. “And I might add on the basis of my 50-plus years of experience in aviation, I think the prospect of large transport aircraft being flown without pilots is highly remote and not something that our crew-force should be worried about in foreseeable future.”


https://www.bloomberg.com/news/artic...r-remote-areas

Viper25

Lots of talk about autonomous airplanes being a threat to jobs. I can see the arguments about how that will be a while. However, what’s everyone’s take on the timeline of single pilot being a threat? Of course cargo first, but how soon?