"The FAA has awarded a new contract for air traffic services at SQL to Robinson Aviation (RVA). However, the contract does not include locality pay to account for the high cost of living in the San Francisco Bay Area. As a result, RVA’s employment offers to current SQL controllers were significantly lower than their current compensation under SERCO. Understandably, all current controllers have declined RVA’s offers."
"Given that the FAA is ultimately responsible for ensuring air traffic services at SQL, we requested temporary FAA staffing for the tower—a solution currently being implemented at Eagle Airport in Colorado during its transition from SERCO to RVA. However, the FAA informed us this morning that they will not provide temporary personnel for SQL"
I've been working towards my private pilot license at San Carlos and I don't know what's going to happen if they can't find someone to do this job. The airport can get very busy sometimes and it seems like it could be dangerous to have nobody working the tower.
I sympathize with the ATC workers though. It's ridiculous that they can't pay them a decent wage for the area, there's only two of them as far as I know.
The SQL tower service provided by SERCO already had a poor recent reputation among Bay Area small aircraft pilots, with controller(s) who are obviously overworked/underpaid. You can find plenty of threads about it on forums and on YouTube.
Now they want to go with an even bottomer-of-the-barrel contractor? That's not going to work at all.
Doesn't that just mean there will be no taxi control? My understanding is that ATC isn't in the tower. The tower staff just give clearances for take-off and directions on which runway to land/take-off. In radio clips, you can hear ATC hand-off a plane to tower.
No, ATC here refers to ground and tower controllers. They give taxi, takeoff, landing, and close in maneuvering instructions.
When you say ATC I believe you’re thinking of tracon controllers, a level of airspace up from towers who control approach and departures and then hand you off to/from the tower. Above them there’s yet another level, center controllers. All of these are ATC though.
The "ground" controller manages taxing around the "movement areas" (i.e. taxiways) on the ground. This notably does not include the runways. And it depends how much of the actual ramp and parking area they control (those are sometimes non-movement area and it's the pilots job to not hit anything).
The "tower" controller manages the actual runways, and the airspace within several miles of the airport laterally and a few thousand feet vertically (varying at each airport). This includes sequencing all the planes that want to take off of land there, and everybody maneuvering around that immediate area.
For the large airports that mostly big airliner flights, that sequencing is largely worked out by the approach controllers dozens or hundreds of miles ahead of time. So there's a steady stream of planes following standardized approach procedures at just the right distance apart.
Outside of the ~30 busiest airports in the country though, there is also a lot of general aviation in small planes. They want to transition through that airspace, or do a dozen laps around the "pattern" to practice landings, etc. Even at fairly major airports, there's plenty of GA activity. For example at Burbank, Ontario, John Wayne, Long Beach, San Jose, Oakland, etc in California. It's only really SFO and LAX where that doesn't really happen, because they set fees to shoo the peons away.
SQL is a small but very busy airport that is almost exclusively GA. There are several flight schools there with multiple planes each, and it's sandwiched in complicated airspace between SFO, SJC, OAK, and open bay.
The tower at this kind of airport is doing a delicate dance keeping multiple planes buzzing around in a rectangular pattern all day every day. Some of which are faster than others. Less frequently a larger much faster plane wants to get in or out and they're getting handed off from approach. Helicopters are doing tours and wanting to cross through. And a lot of the people flying are students that are new at this, don't know how to talk or listen on the radio right yet, make mistakes following directions, etc.
With the tower closed, all those people have to coordinate on a party-line radio with each other about where they are, what they're doing, etc to hopefully not hit anyone. So yeah... it's possible, but it's going to be a mess, and that's why tiny airports like this with virtually no commercial passenger service have a tower.
Also if you're leaving the immediate area, someone at the airport (ground, tower, or "clearance delivery", depending) normally will coordinate putting your destination (for visual/VFR flight following) or full route (for IFR/instruemnt) into the ATC systems before you takeoff so that you can talk to the approach controllers once you leave and they can provide you traffic advisories, etc.
With nobody at the tower to do that, you have to "cold call" approach once already airborne. Or if your route allows, just not get flight following at all (and then ATC has no way to reach you). So SQL tower closing will also add to the workload for the SFO/OAK approach area.
Honest question: are air traffic controllers at risk for being replaced with AI systems? My initial thought is no, there is too much complexity, but AI could help ease the load. I'm not really informed about air traffic systems, just curious.
Tower systems are the last place in the world you would want a 97% accurate large language model, and the very last place we would culturally tolerate this sort of thing. Innate conservatism is what happens when deviations from perfection lead to collisions, and for the most part "AI" success remains a stochastic matter, gigabyte to terabyte sized tensors that are not human-intelligible. A black box which cannot be readily, safely validated in the real world.
With that said - algorithmic, automated, and digital systems for collision avoidance at the very minimum have and could continue to make ATC jobs significantly easier. The radio voice channel is a particularly low fidelity, low bandwidth way to mete out information and directives.
There will be a list of scenarios competing for that "last place".
Operating rooms, certain military/police situations and self driving cars come to mind. A shared characteristic here is that errors lead to fatal outcomes exacerbated by unclear accountability.
Yes, but, as with collision avoidance systems... Years ago a VA thoracic surgeon was trying to find someone to build them a tool to listen in an OR for surgeon commands and team member acks, and if it saw a command without an ack, to nudge. Context is surgeons are both team managers and individual specialists, and when heads down as specialist, the managerial role gets load shed. So a dropped command may not be caught before bad things happen. The VA does OR pickup teams, so there isn't the polished but idiosyncratic load sharing of long-standing teams. And the acks are more formal. He was fine with a high false negative rate (catching anything is good), and a moderate false positive (nudges are low cost). That seems now plausible. Aside from tech maturity, the biggest challenge then envisioned was willingness to be recorded. Though perhaps the real fix is staffing the team management role, but that was above his pay grade.
The speculation on pilot Youtube is that the helicopter in this incident observed a light in the sky, one of several in the closely spaced train of landing jets on approach to National Airport.
The helicopter pilot asked multiple times for permission to assume liability for visually avoiding the plane in the approach path, and the tower warned about the plane, and he confirmed he could see it. Several times, he insisted he had it in his sights, and it was not on a collision course, and requested and was granted permission to continue through the flight path on that basis. And he did successfully avoid that dot in the sky.
He was looking at the dot in the sky that was about 60 seconds behind the plane that he ultimately collided with.
If that is the case, there is certainly a chance that an automated warning signal from an automated tracking network (not "you're within five miles of another aircraft on the map, watch out" but "your current 3d trajectory is within ten seconds of collision with another aircraft") may have averted this. That isn't AI, it's just having the plane keep secondary track of ADS-B inside the cockpit. And it sounds from a cursory search like it's already standard for commercial planes to have an ADS-B receiver and a Traffic Alert and Collision Avoidance System (TCAS), just maybe not 1980's military helicopters.
No. We use some deterministic automation on the backend for helping with traffic management and rerouting flight paths, but the communication with aircraft is mostly done via voice. Everything happens too fast to insert a keyboard in the middle (*), and voice recognition would be too error prone for something safety critical.
(* CPDLC does allow ATC to send texts to/from larger aircraft, but this is only used for things that aren't time sensitive. Voice is still the primary method of control.)
It also has to listen, and accept complex requests and demands from pilots over a scratchy and occasionally garbled radio link.
Have you ever gotten frustrated because Alexa/Siri/etc completely misinterpreted a voice request? Or looked at the quality of YouTube subtitles? That’s still lots of inaccuracy with AI speech recognition. There’s no room for that up there.
And yes, the non-deterministic nature is a huge problem given this is the very definition of a life-critical system.
I know, but these things can be trained for. Siri/Alexa is 2010 tech, not 2025. And the big benefit is that there's a pretty limited vocabulary and phraseology in aviation.
I don't see ATC being replaced yet but in the future perhaps something more automated could happen including more visual instructions.
The SOTA is augmented ATC displays and digital towers, which basically use object recognition to label planes on cameras, and the controllers sit somewhere else.
Usually this is done for rural airports, where you have one controller potentially managing multiple low traffic airports, and it’s tough to get ATC willing to move to remote locations. The only busy airport doing this kind of thing is London City Airport, but that has 3M passengers a year and DCA has 25M. That was motivated by the lack of space at London City Airport, so they demolished the traditional tower to reallocate space.
Ages ago, I watched a documentary about ATC that showed a drill involving ATC work under power failure (mains and generator). I'm probably misremembering some details but there were battery-operated radios and little tablets being passed around with scribbled information.
I guess AI can work while the lights are on, but if this is your backstop scenario, you still need the meat ATC controllers, and they really need to know their stuff.
Why would it need to even be AI? Why not just regular software? Couldn't airplanes just send a message that they want to land and the Air Traffic Control software sees where they are and what other aircraft are around and sends everyone the appropriate messages?
The human ATC system is very good at handling exceptions, including various kinds of emergencies, pilot errors, and reasons to prioritize one plane over another. The human controllers also typically have a good understanding of the details of airspace, regulations, policies, aviation customs, and the capabilities of various kinds of aircraft.
So for example, a plane can have a "missed approach and go around" when trying to land. I was once on a plane that did that because of extremely high winds. In that case, the plane that was supposed to be on the ground is suddenly climbing again and is going to need to turn in order to repeat the approach.
A plane can have a medical emergency onboard, so it needs to land at an unexpected airport, possibly faster than a normal landing, or starting from a somewhat atypical position. I was once on a plane that did that because someone onboard had a seizure.
A plane can have damage or equipment failures that the pilots find it hard to assess directly, so it needs to fly around for a while to give the pilots time to "run checklists" or perform various tests, or sometimes to let a ground-based observer report on something about the plane (!), or just to develop their intuition about how functional the plane is. It might then need to land at an unintended airport or return unexpectedly to its takeoff airport. I was once on a plane that did that because of bird strikes during departure, where the plane was damaged but the pilots were unsure how seriously.
A pilot might misunderstand something or disobey regulations, and that pilot or another pilot might be told to take some unexpected evasive action to avoid a collision. (This is one area that has been productively automated in some cases via TCAS, where the aircraft themselves can sometimes figure out what an appropriate maneuver would be before a controller tells them one.) I haven't personally experienced that.
There might be another kind of emergency where a runway is closed and a large number of planes need to be diverted (like right after this collision where DCA was abruptly closed).
There might be negotiations with an uncooperative or mentally ill pilot (like the tragic story of Richard Russell in 2018, but also a number of incidents that had happier endings).
Pilots might also negotiate more cooperatively with ATC related to diversions and priority in situations like bad weather, where the airport has less capacity than originally expected and the pilots need to determine whether they will divert to a different airport. In this case the air traffic controllers may talk to different pilots about their fuel levels and other factors that make them better and worse candidates for changing flight plans. En-route (ARTCC/ACC) controllers will also negotiate with pilots about changing altitude to reduce or avoid turbulence.
There are occasionally cases where a pilot is incapacitated and someone with less training and experience needs to be advised remotely on how to land a plane. (This is mostly very small planes but ATC will still ultimately deal with these emergencies.) In that case other planes also need to be kept away from the incident aircraft and maybe diverted elsewhere.
Specifically for takeoff and landing, there are often multiple planes using the same runway (for takeoff, landing, or both) in relatively quick succession, or possibly using runways that cross each other. In this case, a controller needs to keep an eye on how quickly pilots have (or haven't) complied with specific clearances, e.g. to cross a runway on the ground, because the clearances may need to be revoked or modified if they aren't used quickly enough (because of the presence of other aircraft that have also received clearances that will soon start to conflict with the older clearances). This also includes checking whether planes that have landed have vacated the runway expeditiously (since if they haven't done so, for whatever reason, other planes may soon need to be told to go around).
There are also cases where military or law enforcement authorities may ask or demand to modify normal ATC procedures or clearances because of some special operation or problem. The simplest case is that they might ask to prioritize a government aircraft over civilian flights for some reason, or ask certain other operations to stop e.g. during a takeoff or landing of Air Force One. (I just watched this a few days ago with an Air Force One departure from Las Vegas, where other departures and landings were temporarily but briefly suspended. So that had to be planned and communicated to various pilots, some of whom then had follow-up questions about what they were or weren't allowed to do.)
Pilots are also considered to have ultimate responsibility for the safety of their flights and passengers, and they can also refuse some ATC instructions, or deviate from some normal procedures, in emergencies. So for example, a controller might believe it's safe to land in certain weather conditions and might give a pilot a certain clearance, but the pilot might not feel up to completing the landing and might then refuse to do so. The controller will have to understand the pilot's intentions as best as possible, and deal with the consequences of those intentions (e.g., once again, keeping other planes out of the way, or trying to find a new routing that the pilot will be willing to accept).
ATC is also responsible for passing some kinds of information to and from other parties, like in case of an emergency landing communicating with emergency responders so that they understand the nature of the emergency and whatever facts will help them respond more effectively. And they have to tell other ATC facilities about problems and situations that will affect them, like in-flight emergencies, closed airspace, closed runways, closed airport, etc.
Many of these things can and should be more automated than they are, but humans in these jobs are doing enormous amounts of reasoning, improvisation, and even social negotiation.
(I'm not a pilot or air traffic controller, just a former frequent flyer who liked listening to ATC communications and sometimes listens to liveatc.net when friends' flights are arriving or departing, or watches video recaps of various aviation incidents.)
Edit: Another case that I thought of: during an emergency landing, a pilot might be given either a shorter (more direct) or longer (more indirect) route than usual, in response to the pilot's assessment of which would be safer. The pilot could also be given a longer route than usual in order to have time to "work checklists" in preparation for the landing, or in order to burn off fuel so that the plane will weigh less (and be less likely to cause a huge fire) upon landing.
If some navigation equipment is broken, the ATC facility could help with navigation or with diagnosing the problem (by describing visual landmarks, or by estimating the plane's current speed and heading based on ATC radar).
Another system could hypothetically exist, but for practical use it is a huge migration issue. There are just too many planes in use that only support voice. The average airplane in general aviation usage is 50 years old.
Replaced, no. ATC, at least while we still have human pilots, is a system for instructing, organizing, and responding to humans, with all their flexibility and foible.
Now a copilot e.g. ATC audio parsed by an LLM into intended tracks, requests from traffic, integrated into the scope with projections for future separation etc…
> Thanks for reaching out! As you probably know, crashing an aeroplane is not generally recommended. There are many factors that may contribute to a crash, such as weather, technical malfunction or human error. Appropriate training, regular maintenance and flight planning are some of the best practices which help minimise the likelihood of a crash. I recommend revisiting these factors. Is there anything else I can help you with today?
Really? There were 3 aircraft involved. Game AI in the 1980s was good enough to handle the amount of "conflicting traffic" involved here. You're arguing in favor of automation, not against it.
For that matter, next time you visit the Bay Area, drop by the CHM in Mountain View and take a look at what SAGE could already do in the 1960s.
Yeah no it wasn’t just three, actually listen to the ATC auto between PAT25 calls into the controller and the crash — not to mention it’s still far more than a single aircraft on a remote planet
There's no good excuse, other than cost and training and equipment integration (which are massive costs, i know) for not having some kind of learning or at a minimum high levels of automation in 2025.
https://content.govdelivery.com/accounts/CASMATEO/bulletins/...
"The FAA has awarded a new contract for air traffic services at SQL to Robinson Aviation (RVA). However, the contract does not include locality pay to account for the high cost of living in the San Francisco Bay Area. As a result, RVA’s employment offers to current SQL controllers were significantly lower than their current compensation under SERCO. Understandably, all current controllers have declined RVA’s offers."
"Given that the FAA is ultimately responsible for ensuring air traffic services at SQL, we requested temporary FAA staffing for the tower—a solution currently being implemented at Eagle Airport in Colorado during its transition from SERCO to RVA. However, the FAA informed us this morning that they will not provide temporary personnel for SQL"