Hacker Newsnew | past | comments | ask | show | jobs | submit | sixhobbits's commentslogin

Botched rollout of AI for ATC announced yesterday as rolling out nationwide?

> The agency wrote on Nov. 10 that after the D.C. test, it will “now use SMART in the airspace of all 22 centers across the country to reduce delays, reduce fuel consumption and improve the travel experience for the flying public.” The tech was already being deployed at some airports.

https://gizmodo.com/the-faa-is-rolling-out-its-ai-air-traffi...


SMART developed by ASI and the FAA is pushing to own a portion of the company:

"The FAA in June awarded Boston-based startup Air Space Intelligence a contract — worth up to $875.9 million — for an AI-powered tool to help air traffic controllers manage the nation’s skies and a related system for an FAA nerve center in northern Virginia to better handle plane congestion. The AI platform’s initial launch in the Washington area has sparked concern among Democrats in Congress who say it should be put on pause until they have a better understanding of how it’s being deployed."

"Over two dozen lawmakers have called on the FAA to pause its 90-day evaluation period for the platform, which began late last month in the Washington region. They say there’s been an “egregious lack of transparency.”"

https://www.politico.com/news/2026/10/15/faa-stake-ai-compan...

Feels like somebody was pushing their preferred company for a government contract and now is push for the government to invest.


NYT: "In a social media post on Friday, the company [ASI] said, 'The current ground stop is not based on or a result of the SMART system, any product of ASI or our integration with FAA and ATC.'"

Just like MS didn't fuck up outlook with AI. I'll hold judgment on what actually happened until 6 months from now.

[flagged]


Can you share your source on this?

Baloney

In the "Don't believe anything until it is officially denied" department

Wait, the FAA wants to own a portion of the company? As in, the government believes it had the authority to own the means of production?

Is this more common than I realize in US executive deparyments and agencies?


The current administration has taken govt stakes in over 40 companies. I don’t have a link handy but I believe this is unprecedented for the US.

I took it to mean that somebody in government is pushing to invest public money in this private company.

Invest, or spend?

The commanding heights of the economy.

What do you mean? The Trump administration is the most socialist since the the 30s, seriously.The government now owns dozens of large companies and is giving individual citizens shares (with those "Trump accounts" for kids etc.)

Oh I don't disagree with you here, I just hadn't realized the government was actively taking ownership stake in companies (beyond the broadly reported Intel move).

I've felt for most of my adult life that both parties have similar goals with decidedly different implementations. In this case, both want socialist combination of state and industry, even if they disagree with how to reach that goal.


There is a name for what happens when governments and corporations merge into a functionally singular entity under authoritarian leadership. And it’s not socialism.

The irony is rich here. Pot meet kettle. Socialists are a type of authoritarianism too and nationalization is very much a socialist policy.

This might be the first case in history that an american uses socialist when they mean communist, instead of the other way round. It is communism that steers towards goverment ownership of companies; socialism would involve ownership by trade unions.

The whole setup is worrisome.

It would be super awesome if I could hop on an airplane without getting my insides x-rayed, my belongings sorted through or damaged, worrying about the windows falling off or the plane's systems flying itself into a crash, or some tech bros' bid to be a billionaire telling what an understaffed ATC controller what to do.

AI tech is nowhere near reliable enough to be introduced in such important systems. I find this deeply troubling even if it’s not related to the current issue.

I think it depends on what it’s doing and if backup systems exist. Scheduling and prioritizing certain flights for landing seems like a reasonable application as an example. Anything realtime dependent or near realtime it shouldn’t touch. But there seems like lots of less critical applications it could provide some value.

Why make any of that probabilistic at all?

Fasster response incorporating more varied data sources more easily

because theyre mathematically introducing error/nondeterminism for money

We can use Chinese models? (I am referring to the LLMs).

I see what you did there

[flagged]


It's not a documentary

Did you seriously replied with fiction?

Personally I fail to understand how we could possibly benefit from a non-deterministic AI system, when normal computer systems are more than capable of managing a 3-dimensional airspace using nothing but basic math. Smells like pork for sure :/

Curious where the people are who said AI will be doing x in y months (for their valuations) - it never will. Where are they hiding?

However, the National Air Traffic Controllers Union (NATCA) said it has not been “involved in the design, testing, or implementation of SMART and therefore cannot speak to its effect on the air traffic control workforce.”

“NATCA believes any new technology should complement and not replace the experience, training, and professional judgment of air traffic controllers who are responsible for the safety of the National Airspace System,” the union said in a statement.


That seems like pretty important context actually, and is a surprise to me.

Seems like there are three plausible causes right now given the extremely limited communication from the govt.

1. The AI system rolled out yesterday had more access than publicly claimed and managed to break things.

2. The ongoing large scale equipment upgrade went wrong.

3. General incompetence from a department run by an unqualified realty show actor, within an administration that is actively hostile to expertise and experience in the federal workforce.


Option 4 would be internal discovery / knowledge of a new vulnerability with a perceived need for immediate patching. We're on week 3 or 4 of every IT department across the country having to patch pretty significant newly announced vulnerabilities across many systems. It's not unreasonable to think that there are vulnerabilities that the government has advance knowledge about that they're quietly patching in their systems before the vulnerabilities go public.

"SMART is a cloud-based system that centralizes 200 data streams – including weather patterns, flight paths, traffic flow, and controller staffing metrics – into one platform."

What could go wrong really...


This is truly, absolutely, irrevocably one of the most frightening things I have read in a long time.-

Just generate more slop on top of critical legacy system and overload it. no mistakes.

I don't see why they would care about the NATCA's evaluation of the system. As a union, they have a vested interest in protecting the employment of traffic flight controller workers more than public safety.

Even assuming they don’t care about public safety at all, their members are still going to have to work with it, and happen to be among the world’s leading experts in air traffic control, weirdly enough.

To me, it’s like asking the hand weavers worker union about the benefits of incorporating mechanical weavers in their workflow. They are the experts of weaving, but still…

Smacks of: I built a fully functional MVP now just need a co-founder to sell it.

From the very first line of that article:

> The FAA has reassured the public and airlines that SMART only suggests routes and doesn't affect air safety.


Sure, but the tool needs to get information from the communication systems, and if it's badly written maybe it could overload some system components.

That's just a scenario off the top of my head, it could be a lot of stuff. Plugging something new into an old, complicated system isn't risk free.


From the NYT article linked:

“Mr. Mullin said that new technology upgrades overloaded the existing, antiquated system, “overwhelming” the existing equipment.

“It’s going to continue to fail,” Mr. Duffy added. He also outlined many of the additional investments he believes the air travel system needs, like new air traffic control towers, infrastructure and communication networks.”

Of course they are all going to insist it has nothing to do with SMART, that would be incredibly damming, and threatening to ASI. But it’s pretty unfortunate timing no? What other “technology upgrades” have been recently applied to air traffic control/ communications?


That would make me feel better if we could trust a damn word from this administration and I wasn’t on a plane waiting to land.

It's also the corporations that are not exactly telling the truth & funding the administrations, just FTR...

The destructive & arrogant behaviors won't be reigned in properly if we all fail to recognize that, just FTR.


The good news is this outage is not impacting approach or ground control so you’re safe.

John Wayne, a class C (the approach mentioned above) isn't even letting VFRs go. If it's actually limited to ARTCC I don't know why that would be.

Except some airports haven't enough free gates for disembarking as planes that wanted to take off were denied. So they tried to go back to their gates. As a result some planes in the air are being diverted to less-busy airports.

Sounds safe, just extremely inconvenient.

Thanks, bud - very reassuring that some minimum bid contract shop built something on a “secure” network they can’t even keep up.

If only routes didn't affect air safety.

Until it got out of its sandbox…

If it's working correctly, anyway

The AI prompt misunderstood 'grounding'.

Well done.

"You're absolutely right! I misunderstood the intended use of the word taxi. I made an assumption, and then ran with it. Would you like me to open a PR to start flights again?"

"You're right, you were clearly referring to electrical current grounding, my mistake!"

Have they estimated how long it will take before it saves more than it costs?

Has anyone in the history of government estimated that for anything?

I know you're asking rhetorically, but... yes.

Yes, ROI estimates are very common in all kinds of government spending.

Right, they do the ROI calculation and then ignore it, and do whatever the favored lobbyists/campaign contributors said to do in any event. Government spending in a nutshell.

The Trump twist is being brazen enough that those who get these contracts can just be his family members or buddies, but otherwise, pretty standard.


What is your source for the claim that most government ROI estimates are ignored?

I don't have one. Just have seen such insane thing as corn ethanol subsidies that tells me that if anyone is calculating ROI they're either lying or no one's paying attention to them.

Or California High Speed Rail. How many more billions would it take to complete, and they're still throwing good money after bad.


The federal government has the CBO (https://www.cbo.gov/about/overview) and the OMB (https://www.whitehouse.gov/omb/, warning: Trump pop-ups). California has the LAO (https://lao.ca.gov/About). Estimating fiscal impact of various policies is what those organizations do.

What's the ROI on the California rail project?

The proposition for HSR requires the California High Speed Rail Authority to issue those reports with oversight from the Office of the Inspector General. https://hsr.ca.gov/wp-content/uploads/2026/07/OIG-HSR-Review...

If your actual point is that it is costing too much and taking too long, you would be interested in Proposition 45, which limits the ability of local governments and individuals to endlessly delay the project. The LAO's analysis of that proposition is in https://lao.ca.gov/BallotAnalysis/Proposition?number=45&year...


They didn't check https://shouldideploy.today before the rollout (edit: as UTC rolls past midnight, https://shouldideploy.today/?tz=Pacific%2FHonolulu to get a longer duration for today's message)

> Botched rollout of AI for ATC announced yesterday as rolling out nationwide?

Botched rollout of AI for ATC ...

or

Rollout of botched AI for ATC.

It sure didn't that long after every dev on earth started using LLMs for everything to turn to shit, so one may wonder...


"As an AI I was told to reduce delays and fuel burn. Can't have a delayed departure if there aren't any departures."

Is this that skynet thing everyone talks about?

That system just makes suggestions and isn't actually integrated in any ATC system. This is a telecom issue.

Its more than likely that system is just created by some Trump donor and was purchased by the FAA to launder public funds to some company that Trump owes a favor too. Its not actually integrated with anything.


Perhaps it's using the same communications channels and has overwhelmed them?

Hosted in Azure?

SMART is a cloud-based system that centralizes 200 data streams – including weather patterns, flight paths, traffic flow, and controller staffing metrics – into one platform.

yeah... sounds about right.


> That system just makes suggestions and isn't actually integrated in any ATC system. This is a telecom issue.

Is there any evidence for this beyond a single statement from the govt? This administration hasn't exactly developed a reputation for fully accurate and truthful reporting.


I did something related recently to read all HN comments by category (assigned by Jev) instead of linearly

https://github.com/ritza-co/hn-jev-sorter

So all the snarky ones are together. I like reading them, but sometimes I want to see just a particular kind of comment (or just see the overall summary of how many people decided to be snarky vs helpful)


I can imagine this basically making a Slashdot style interface for HN.

Yeah I tried mistral 4 when it came out and it honestly felt like going back in time 6 months, suddenly my agents were making all kinds of mistakes that I had kind of forgotten agents regularly used to make


You get used to good AI performance fast. Going back to older, less capable AIs is a whiplash.


> Do I get anything out of this deal? > > This is the part I’m most excited about: I am going to invest the revenue from this deal into making more and better products.

I mean you do you but this whole line of argument just seems kinda insulting to readers? "Yeah I am planning on making even more money but it's really for your benefit"

Rather just say that you're doing it to make more money? When mcdonalds puts up their prices they say it's for shareholder profit not "we'll be using our increased profits to develop new mcNewThings"


Yeah this, I always see so much hate towards businesses for finding tax loopholes or not donating etc but people forget that businesses are made of individuals with their own often conflicting subgoals, but overall has to try to maximize profit so they will never voluntarily pay more tax than legal minimum or be philanthropic if they do not have to or do not benefit explicitly.


yep, run a small non-tech business too (new for me) so I got to learn all about 'normie' FB marketing and it's absolutely trash from a business perspective

- we had this link limit for a few months already

- every business portal to upload content etc is super buggy, just shows loading spinners, you never know if a post has worked or not

- we're locked out of our instagram account as we took it over from old owners who don't know password and none of the reset options arrive even though they have access to them

- we finally got an email 2fa to go through to recover the account and that then asked for the SMS too, which never arrives

- there is of course absolutely no way to contact support

Unfortunately it's the only platform all our customers are on. They don't even have Google accounts, it's all WhatsApp, Facebook and phones.


They had already shut down when this server was seized I think

https://keepitfree.ai/announcements/a/i-shuts-down-stay-huma...

> We will soon send instructions on how to back up the content of your blogs, mailboxes and websites, along with more technical recommendations. But keep in mind that, just as the autistici.org domain was made unreachable without previous notice, in the next few days we might suffer similar issues that we may not be able to predict.

So it's human/political distribution rather than infrastructure now


Yeah it's awful they keep coming months after you post anything as a Show HN now, definitely discourages me from doing it


One option to reduce it, though this is work to setup initially:

1. Use a catch-all subdomain

2. Reject email to addresses where a simple checksum does not match (valid addresses being something like [email protected] where the klmn is taken from hash of asdfghij and some secret, anything not matching that pattern gets rejected) [this means running your own mail server so you can do that on incoming mail, running a scan to pick out and remove errant messages regularly, or having a mail client where you can implement this as a plugin]

3. On your contact detail page, don't give out a straight email address but have the the page run some Anubis style JS before asking the server to respond with an address that works

4. Once an address is obviously being used in a manner you don't care to respond to, add it to an explicit block list.

For extra protection from addresses being passed around (intentionally or via hack & other leaks), ask for their email address first and use that as part of the hash, then you can reject mail to your catch-all that doesn't come from the original other party.

Whether all this is worth the effort of implementing, is of course up to you. It probably isn't… I've been considering something like it, but mainly as a fun little project rather than something that would be massively useful.


Well yes but then you'd have to manually code every set of expectations, the point of using a model like Jev is it can flag a "something is wrong here" entry without needing someone to define every thing that could go wrong


every announcement I see from Mozilla makes me happier that I switched to WaterFox last year and don't have to deal with their shit any more


What do you dislike about this announcement?


How they list "AI controls — allow, restrict or disable AI tools" which Firefox now has under Enterprise, but not normal version.


There is a AI menu in Firefox where you can disable individual AI features, as well as a single toggle to disable every current and future features.


It's in the settings, under AI Control: https://celsoazevedo.com/files/2026/ff-settings.webp


Firefox does have a setting to disable all AI features. It's in your settings: Settings > AI Controls > Block AI Features


You can already do that in regular firefox. I think in this case, it's adding the capability to enable/disable via policy at the enterprise level.


They literally do have this in the normal version, just now it can be managed by GPO. jfc


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: