The Future Of Artificial Intelligence
 

I'm a student at a part 141 school and I have been doing a ton of research on automation in the cockpit, more specifically artificial intelligence.


My question's for the pilot community is this:


1. What advancements will we see being implemented in the cockpit at a part 121 carrier in the next 1-10 years... (what technology is already semi close to being done with development/deployment.)

2. When will AI kick out the co-pilot/pilot monitoring in the cockpit. If you don't believe this please articulate your point because I feel this is inevitable in the not so long future (20-30 years my guess)

3. When will 121 carriers try to automate the entire process of flying? I know what you're thinking....

"but the public would never trust a computer to fly us around" I personally believe once we go fully autonomous in the automotive industry and people are accustomed to trusting AI with they're lives to transport them in cars it will change the perspective of the public. (not saying that automotive AI and aviation AI are at all same level of complexity, but I do think public opinion will change)


I know I'm asking you all to whip out your crystal ball, but I think it's a conversation this community should discuss. Plus it would be great to hear from some guys that have been are currently in the aviation industry.


Thanks your all your opinions.


- John

My guess is the answer to question 1 will be 20-30 years, in cargo aircraft. Then when they start crashing and scaring the public the answers to 2,3 and 4 will be never in anyone here's lifetime.

I do see advances in automation, next gen ATC system, more efficient routes, but the two person cockpit is here to stay for a while, being in a 141 program you should be thankful.

We'll see single pilot within a decade, fully autonomous aircraft, and cars will never happen unless quantum computing becomes a thing.

Scoff. There's absolutely NO WAY this would happen without HUGE infrastructure changes to ATC as well as aircraft systems and pilot interface. Certainly not anywhere close to happening in the next 10 years.

Before we go down this rabbit hole, let's keep in mind we are still dealing with the FAA. The same organization that still will not allow quick-don O2 masks to suffice when one of the two pilots leaves the flight deck. So, 20-30 years from now if we're down to one pilot, is he gonna be sucking rubber for the whole flight? :rolleyes:

1. What advancements will we see being implemented in the cockpit at a part 121 carrier in the next 1-10 years... (what technology is already semi close to being done with development/deployment.)
We already use CPDLC, ADS-B and DCL to receive ATC clearances and communicate directly with ATC via text and in some cases FMS interface. As I'm sure you've already found in your research, moving away from an ATC system reliant on voice communication is the natural goal of this technology as it continues to develop. Personally, I think that will be the next major leap in technology in the commercial aviation industry.

If you think about it, most of the potential automation you discuss later in your post hinges on this happening. Until ATC has the ability to issue instructions using a feedback loop that isn't reliant on VHF voice, removing pilots from the equation has no chance of happening.

2. When will AI kick out the co-pilot/pilot monitoring in the cockpit. If you don't believe this please articulate your point because I feel this is inevitable in the not so long future (20-30 years my guess)
As I just said, the first thing that must happen is ATC modernizes. One of the myriad of reasons we have two pilots is a cross-check of spoken ATC instructions. AI isn't going to be an effective PM in a busy NYC approach environment, isn't going to catch read-back errors or be able to utilize CRM principles we currently count on to capture and correct human errors in the cockpit and in front of the ATC radar scope.

The second thing that will need to change is aircraft systems. No current production airliners are made or certified to be flown by just one pilot. That's going to require a major re-design of future cockpits. In order to have just one human pilot on board, there must be major changes made to deal with the potential for his incapacitation. In reality, the human pilot must now become the back-up to the AI since a human-centric system no longer makes sense when allowing for incapacitation. Maybe that'll finally relieve the requirement for constant O2 masks.:D Current human controlled inputs to normal things like altitude limits, assigned headings, actuation of flaps, speed brakes and landing gear, nosewheel steering, lights, etc. will all need to be automated. All the human responses to non-normal events using fire handles, fuel cut-off switches, CSD disconnects, cargo fire discharge switches, etc. will also need to be re-designed to allow remote operation. Basically, until the entire aircraft can be remotely operated from a ground station, the option to remove one pilot does not exist. While the technology to do this certainly exists right now, the incorporation of that same technology into a certified commercial airliner or freighter is not right around the corner.

3. When will 121 carriers try to automate the entire process of flying? I know what you're thinking....
See my response to #2. It's not the two step process you're making it out to be. It's not baby-step #1, remove one pilot and see how that goes before moving on to removing both. It's an all or nothing proposition. Removing one pilot is still going to require the aircraft to be completely automated to allow for the possibility that the remaining pilot is no longer available. But, the first step still requires a human pilot station, while the second step does not. So, does that mean they re-design the aircraft and completely remove it or leave it in for an occasional human pilot.

The ground control infrastructure required to support a pilot-less airliner is another vague but major re-design necessary. I don't think this stretches the imagination or technology near as far as some of the airborne requirements. However, modifying the nation's major airports to deal with a bunch of AI controlled aircraft moving from gates to runways and vice-versa isn't a minor task. Leave one pilot in the loop and it's nowhere near as big a change, if any at all.

The other factor in the public maybe (a very big maybe) eventually accepting AI airliners is system failures and in-flight emergencies. Though relatively infrequent, when you look back on some of the major events that have been successfully handled by human flight crews, it's not a given that the outcome would have been similar if AI was completely responsible. Human interpretation of the circumstances, manually flying the aircraft in some cases (US Airways 1549, UA 232 or the Gimli Glider for example) is an intangible, valuable contribution that can't necessarily be replicated by AI. Sure, human involvement also causes some accidents that AI might not. But, I don't think the traveling public will ever be okay with a crash that came as a result of HAL getting to the end of his command string of possible options because his programming lacks the dynamic problem solving abilities of the human brain.

Personally, I can imagine a one pilot cockpit sometime in the distant future, guessing 20+ years at the earliest. Lots will need to change and be changed intelligently. It'll probably happen in the freighters first. I seriously doubt we'll ever see a human-less, AI controlled passenger airliner using known or realistically forecasted aviation technology. That possibility is so remote and so far off into some future aviation system that it's pure speculation. No average paying passenger is going to be ready to accept that for a long, long time.

That's a future that's even more scary, because if you still have two pilots in the next round of automation, pay is going way way down.

I think we are asking the wrong question here. It's not whether automation will replace humans, but will automation replace skilled labor?

If planes become so easy to fly, you can replace a pilot in two weeks, no reason to have pay rates over $50K.

Current automation is about as advanced as it's going to get within the existing system we have in place. Some systems on the 777 are intentionally less automated to keep the humans involved. The "next round of automation", if it happens, would be what I described to get us to one pilot. That's not going to happen just because it can while keeping two pilots. You're forgetting about the $$$$. Boeing or Airbus are not going to make the massive capital investment required to advance current aircraft to one pilot unless airlines are going to buy them. Airlines aren't going to order them without regulatory changes complete allowing single pilot and if they do order them, they're certainly not going to do that only to put two pilots in them. This process of R&D, field testing, regulatory changes and ATC/industry infrastructure modernization is going to be glacially slow.

:confused: Um....they're already easy to fly as long as the automation works. The reason we're there now is to handle the non-automated stuff and most importantly for judgement, problem solving and manual flying skills when things don't work as advertised. Further automation isn't going to remove the potential for system failures and the requirement for human intervention.

Describe this aircraft advancement that would allow a person with two weeks training to deal with a malfunctioning autopilot and be able to take 3/4 of a million pounds of airplane, fuel, cargo and people and hand fly it back to a runway safely after losing an engine.

One pilot, with the required changes to airplanes and regs...... I agree with you, 30 years.

Zero pilots? Not going to happen, at least while there's a 2 and a 0 as the first two numbers in the year.

In our litigious era, underwriters will insist on seeing years of safe operations (perhaps by the military), before writing insurance policies. Without that, airlines will refuse to go along.

I'd agree with most here until I read the Norwegians are on the brink of AI shipping--no crews. But, I think the expense of AI aviation won't make crews obsolete; we're less expensive than the capital cost of autonomous planes now. Not to say, technology won't make flying even more boring than it us now. At least the cruise portion.

GF