Honest question, I had the same issue when my work wanted me to stick a chatbot in front of our HR portal and I never resolved it to my satisfaction.
You have to measure that problem against the status quo, which is that current government websites are very hard to understand and leave people with very little idea of what’s going on.
I France service-public.gouv.fr is amazing
https://www.service-public.gouv.fr/particuliers/vosdroits/F1...
I was recently back to Europe and it’s basically comedy (tragedy to be precise) how bureaucratic the whole system is.
Humans also make shit up all the time.
If an LLM makes shit up, your one and only recourse is "computer says no".
Kind of like saying something happens "all the time" and sharing a single example from 15 years ago. Not to mention that a private company's HR team is not the government.
I see it every day in the text, images, and video posted by AI proponents on social media. Full of errors and wtfs and the people posting the tripe don't seem to notice it.
Maybe you've just stopped paying attention so you don't notice as much anymore?
Okay. So why should government websites like the French one be well designed? Maybe it should not be.
At the bottom you also have other relevant links (usually to the website of the specific administrations of what you want to do) including reference to the current law that the page is based on (https://legifrance.gouv.fr is also great)
For what I used it for, it was never ambiguous in determining your situation
Honestly, using an LLM the way america.gov does it, is pretty much the same thing as using search engines, except the result you wanted is almost always in the first result, not 2/3 of the way down the 4th page. Or 42nd on a list of things you are not interested in at all. So much more efficient!
Of course, we also paid for local calls so things never got as popular as they were in the US.
To enhance the status quo you don’t need LLMs, you need people who care.
It’s not that they can’t read anything at all, it’s that they require more basic words. A chatbot is specifically the perfect thing to keep explaining something in more and more specific and basic ways as necessary.
And I agree but your comment, IMO, reads accusatory. If that's not the case no worries.
> It’s not that they can’t read anything at all, it’s that they require more basic words. A chatbot is specifically the perfect thing to keep explaining something in more and more specific and basic ways as necessary.
I mean, sure, if it's accurate. Given these things' propensity to just make shit up, as stated, that feels like a way to make the problem worse, if anything.
Edit: And again it feels like a missing necessary component to this is: if a user of this LLM is told by that LLM that it's legal to, I dunno, dump waste oil in a drainage ditch, then that statement needs to be treated as authoritative and it's no longer appropriate to ticket the individual when they get caught doing that. It's similar in my mind to car dealers deploying these stupid things and some people managing to get them to agree to sell them a Toyota Tundra for $200. If you're going to put them in a place where people can reasonably assume "this looks like a proper avenue of communication" then what it says needs to be binding.
And if it can't be, then don't use it.
I think the major concern I’m picking up is that the potential for misinformation is high, though maybe we can agree it won’t necessarily be common. One person might get told to pour it in a ditch, but statistically most will be told to dispose of it properly. I also don’t think “the LLM made me do it” will work, or at least not for long, and at least not for companies. Maybe an elderly owner could get away with a fine and a reminder not to do it again, but if you’re big enough for lawyers (and thus big enough to really do damage), you’ll be expected to know better. Judges ARE still humans!
When you need to work with the government, the last thing you want is a program limited by a specific set of rules and interfaces. You want a person who understands your (sometimes unique) issue, who knows how things actually work, and can work around the restrictions imposed upon programs. Someone who can pick up a phone, help you out of bureaucratic corners.
America.gov AI: Call the Patent Electronic Business Center for USPTO.gov account and customer-number association issues. Toll-free: 866-217-9197
Not bad at all!
If your salary is 40k euros with no advancement in career and keep being threatened you would be replaced with AI, I highly doubt anyone would care.
> To enhance the status quo you don’t need LLMs, you need people who care.
they can be taught where they went wrong.
if a pattern emerges they can be moved to a role more fitting for them. or removed from a project entirely.
we can judge their effectiveness from past performance.
we can put them in less important roles and gauge whether or not they should be moved up.
so, no, pretending that the correct route forward is to “remove humans from the equation” for a bot that doesn’t learn, is never held accountable, and not held to the same standard as humans is silly tier thinking.
bad bots can be retrained or shut down far more easily
particularly most of the commercial models.
again, i’m sure we can all come up with a thousand “hypotheticals”, and tbh, the hypothetical parade is not something i’m interested in having. but i’ll state again, “entirely removing humans” from the situation is silly-tier thinking.
particularly as even the ceos of sota frontier producing models will each and every single one tell you to never trust their model.
“our model is smartest thing in the world…”
…next breath..
“wait, you trusted our model? that was silly of you. always double check it”
I don't see such a path with an llm.
This seems like a great source for domain-specific RL, but even without that, we can expect there to be higher-quality LLMs that can be trivially swapped in within a year, if not a week.
i didn’t say this at all. did you respond to the wrong comment?
i was responding to the comment which said:
> … remove humans from the equation, then.
> We shouldn't use an LLM because...LLMs never improve?
the context of my response was “removing humans from the equation” is sillytier thinking.
not “We should never use LLMs”
Maybe a human should be available as a backup or something, but not because LLMs don't "learn" -- the system seems to learn in all relevant senses.
(I'd also object to the no accountability, but another subthread is already on that.)
As an individual, yes. As a population, no. They elected a conman twice and would rather burn down the country rather than changing their view.
Saying they didn’t learn is an incorrect analysis, since they learned well, based on the inputs they were provided.
I always recommend Network Propaganda for an empirical analysis of what the patterns actually are.
I used to believe Trump winning the 2016 election was a product of rightwing propaganda and fringe theory, but with the 2024 election, I had to revise my belief. What if this is the mainstream mindset of Americans all along, and Trump simply tried to capitalize on that?
The later chapters are where they go into the more recent details. There is a video of one of the authors (not either Robert Faris or Hal Roberts) is giving a talk at a university panel, and he gives a shorter synopsis of the main mechanism, Its about 11 to 15 minutes in.
The set up of the book puts the evidence for the case they make later, namely that the left and center media are trying to follow journalistic standards, while the machine on the right is entirely different.
I'm pulling from memmory here, bu as an example - a fringe theory will start circulating on the fringes of the internet, the most engaging narrative will be brought up on the podcast circuit. At some point a talking head (Guiliani back in the day) will reference the theory on the news. Which will in turn be referenced by the government who will say this is a topic being discussed on the news.
The other distinction they make is that people on the right who say the narrative is wrong, simply don't get platformed.
They go through this in exhaustive detail. If you know most of the history you can see if skipping to later sections works for you. Its the best description of the actual media mechanics at play.
The 2024 elections were the ones that forced me to really re-examine my priors, and I came to a similar conclusion that Benkler and co did. Their book simply brings all the receipts.
I was referring to the government side, as a mild quip.
Isn’t that one of the biggest reasons why an average user gives up and closes the tab?
The US government, on the other hand ranks close to the very bottom of 1st world nations for forms culture. Forms that are difficult to read, and even more difficult to fill out accurately, often requiring supplementary information from multiple sources. So much so that people are advised to consult a lawyer just to fill out an application for Tax Identification Number! (My personal worst experience with American forms culture).
So I don't think it's entirely fair to compare UK government websites to American government websites. Navigating a bureaucracy that's fundamentally broken is a significantly different kind of problem.
fwiw, I'm Canadian. Canadian forms culture falls somewhere in between. There aren't a lot of forms that are terribly difficult to fill out; it's rare to find a form that requires more than about a dozen pages of supplementary information (all of which comes with the form when you download it). But still not anywhere as easy to use as UK forms are.
If America was going to follow China’s footsteps, then what was all the hullabaloo about freedom and democracy all about.
Even if we grant that there is a difference between facts about history and … recent history, the status quo is improved by building better websites.
A website is also a document of regulations and a source of evidence. If a government LLM hallucinates a new feature or rule, and someone acts on it, a fresh legal hell has been created.
America is also unique, in that one part has part of its strategy to make the government as ineffective as possible, since it benefits Republican political goals.
The solution to that is to make the sites easier to use.
If only there was an office that cross-agency allowed for a consistent and reasonable design of all public pages and that they are accessible... like the uk https://www.reddit.com/r/AskUK/comments/17s14k4/how_did_we_e...
I don't think we have a good answer for liability here, like for Tesla FSD, Waymo, or even the current frontier models.
People don't read disclaimers nor warnings about the content, but usually won't come to the HR to complain a mascot fed them weird info they didn't bother to check.
Compare this with the benchmark.
Earlier this year I called ABC with a question about my liquor license and spoke to three different agents. Each one gave me a different answer. All of them were wrong.
There's a ton of hearsay and rumor about how the US Government works, even among educated/sophisticated people who are dextrous with bureaucracy. The problem gets much worse as you go down the educational ladder, which is most of the US population.
The Perfect, please meet The Good. You are not enemies.
If you're really curious, I want to get a Type-86 tasting license for my store. The ABC site[1] says my kind of store (<5000sqft) is "generally" not qualified, which makes it sound like there are exceptions (spoiler: there are).
The first agent told me that the license is not an add-on, meaning I'd have to complete a very long and expensive process to obtain the license with the risk of getting denied. This was wrong. It's an add-on.
The second agent told me that I was not eligible because I have less than 5000sqft of space. Not true (see below).
The third told my wife that 75% of the items in our store need to be alcoholic beverages to qualify. Wrong again.
I finally asked my legal LLM, which pointed me to the statute, which says, "Type-86 ... can be issued to businesses which also hold off-sale retail licenses [that] earn at least 75 percent of their total gross sales from the sale of alcoholic beverages." Ultimately, it comes down to how ABC interprets and enforces this law, so I really expected to get some clarity from them rather than complete falsities.
[1]: https://www.abc.ca.gov/instructional-tasting-license-for-off...
The bar for something like this should not be 100% accuracy. It should be 100% accountability and transparency, followed by being at least as accurate as google search.
Similar to autonomous driving: it doesn't need to be perfect, just less likely than humans are at causing an accident.
"Sometimes computers just make shit up now but that's OK because humans do to and if they do we can just sue them" should not be acceptable.
That isn't how LLMs work. LLMs are statistical language models, not search engines[0,1]. They can be prompted to call external software to search databases, but they themselves are not capable of doing so, and more often than not they generate responses based on their own model, which may not be accurate. We've had technology that was capable of searching databases for decades without the quirk of not being capable of presenting that information accurately.
>Are you saying LLM's are not suited for searching through text?
I am saying that first and foremost they don't do that and furthermore that they are less suited as a substitute for that than what we had before. An obvious example of this is the AI feature of Google Search, which I've seen hallucinate results numerous times. But you can also look up the numerous times AI has fabricated citations when used in scientific research.
[0]https://medium.com/@himadri.abm/large-language-models-are-no...
My local county government has a pretty good website, but even then it’s completely overwhelming if you try to find something off the happy path of the most common services/tasks. I can’t imagine the nightmare of trying to organize a federal portal manually and keep it up to date. Plain search isn’t sufficient because there are so many overlapping functions that are just slightly different.
I hate to say it, but this is one area that an LLM assistant actually makes sense. Maybe it needs a second validation pass with routing to a human assistant if it can’t figure out how to give an accurate answer to a query.
(Personally I’ve found that “search assistant” LLMs are the only task where I’ve found LLMs to be occasionally helpful to me.)
Edit: the big problem with this sort of thing, even if not AI powered, is that government itself isn’t well-organized and the legally “accurate” answer may not be the correct one, as per this comment: https://news.ycombinator.com/item?id=49895741
https://insidegovuk.blog.gov.uk/2026/03/16/5-things-we-learn...
So the question is is the chat bot better than people trying to understand complicated government websites by themselves?
No government employee is your lawyer. If the government gives you incomplete or wrong information, you are still liable for not doing the correct thing.
Having a website that does a best guess at what steps a person must take after changing a name or getting married is better than nothing, but the status quo isn't "nothing".
It needs to be kept up-to-the-minute updated and it can't skip any nuance / details. Does anyone here really trust the lackeys who spectacularly failed at DOGE to do boring and highly detailed work?
My team experimented with this a couple of years ago with a complex government regulatory body of knowledge.
Google and Microsoft have APIs that will ground with and reference the specific policy documents, which helps. Even then, it was more useful as a job aid for people already knowledgeable of the data than for a layman.
With all LLMs, the quality of the question correlates with the quality of the answer.
Meh, the kind of person who would accept bot output without even reading its sources are already trusting the world's dumbest model, Google's "AI Overview" by searching and reading that instead of clicking any results. If we can get even some of them to start out at america.gov instead of google.com for those questions, it's likely going to be far higher quality due to using a better model and having been trained, I assume, to only answer using knowledge from a .gov primary source rather than guessing based on vibes, or on jokes once seen on Reddit, as AI Overview tends to.
Doesn't really have any bearing on a similar system for HR. The government is sort of unique in a bunch of ways, but not really being responsible for what their low level agents say is one of them.
I'm not sure how to fix that, to be honest. I'm also not sure how much worse it is than trying to Google it? Google is full of stuff that's either wrong, or won't apply for a reason that takes some reading comprehension to grasp (e.g. state-specific rules/programs/etc).
Someone mentioned a comparison to Obama's healthcare.gov rollout; that site had a 1% success rate during the first week it went live so it's safe to assume there have been some lessons learned since then.
The main lesson learned was that it was ineffective to rely entirely on outsourcing. The government needs top-notch engineering talent in-house. This led to the creation of the United States Digital Service (USDS), a fantastic group of people which did some of the best web design and engineering on the planet.
Trump killed them, of course, and the website is now entitled "United States DOGE Service". I hope that, some day, I will no longer be embarrassed to be an American.
Seriously, just look at this crap. Fonts so large they become hard to read blurring into existence as I scroll, giant autoplaying video chewing up my bandwidth and CPU, giant useless pictures, a "food pyramid" even harder to read than the one from the '90s, low-contrast colors, not a citation in sight, one picture too small to read that can be expanded by clicking on it with no indication of this interaction, and it looks nothing like any other government website in existence. The total page weight is over 30 megs. Worthless at conveying information, but it looks capital-M Modern (and gives me a capital-M Migraine).
https://wave.webaim.org/report#/realfood.gov
(The information is crap, too, but you already knew that.)
GSA's work was mainly done by 18F. Which was also destroyed by Musk and pals at the same time they destroyed USDS.
What made USDS (and 18F) special was not merely having web developers in the executive branch. It was that they were among the best in the world, and drove standardization and improvement across the entire government. Now we've got AI bros with buckets of eye-catching (eye-melting) slop.
Clarity is important for things like filing your taxes or applying for a visa, and even then you don't need a boring design to do the job. All it takes is serious amounts of UX testing done by someone with real experience.
But that's kind of irrelevant. This is a government website. A >30 MiB download; poor support for reader mode, custom CSS, or screen readers; excessive animation; and low contrast are all objectively poor for accessibility. This is why the people in 18F and USDS made websites that look different from private industry: the goal is to provide essential services and information to everyone, not to make a real impression on the subset of the population that can use the site.
If the website isn't important enough to design it to be accessible by every American, it isn't important enough to spend my money developing it.
What on earth gave you that idea? This admin has made it very plain whom they serve and who must serve them.
That being said, I’m aware of a vulnerability in that regard of it is attending to guardrail answers to the .gov and .mil domains.
I kind of think we need to expect more from our governments. If this was just a business I wouldn't mind so much. But this is a government and it has to serve everyone, including people who are mentally incompetent, an it has to do that fairly. All of us, the entire American citizenry.
So rolling this out is incredibly irresponsible in my view.
The first suggestion I saw was "I just got married, how do I change my last name?"
It tells you to update your Social Security info first and gives you some helpful links.
ChatGPT tells me the same things with some of the same links. It didn't even have to ask if I was in America because it found my state based off my IP I assume.
2. US won't run their own local model they simply use one of the three who's the coziest with the government paying it $$$ and user data. And there was no request for proposals or public tender.
From the privacy policy[1], "Our AI providers operate under Zero Data Retention (ZDR) agreements and retain none of your prompts or the responses they generate. America.gov’s own response cache is separate and is described in Section 5.
Our AI providers are contractually prohibited from selling your information, building an advertising profile, or training their AI models with your data."
On the other hand this is just going through one or more of the big AI companies anyway. So they still get data from users, now we're just funding them with our taxes instead of using one of many LLM interfaces available for free.
The privacy policy actually looks OK though, I'll be interested to see what the EFF has to say.
It’s likely just a mask of either ChatGPT and probably through some massively bloated, nepotistic federal contract. There may be some additional guidelines and instructions but it’s probably not RAG on anything that isn’t publicly slay accessible to any other model, even if it is hard limited to federal sites/domains.
In addition to what others have said, ChatGPT is also available free and without an account I believe. What are you more likely to think of when you have a question, your AI service you already use and have your information in and maybe even has access to your documents, or are you going to remember to go to America.gov for basic chat AI?
Frankly, it would be really interesting to benchmark it against the other models when it comes to accomplishing things related to the government, i.e. Which information is more accurate, useful, and actionable.
There’s also another matter I find a bit troubling depending on the nature of how this came about, it could also be something that creates an inextricable incumbency in whichever AI company is behind it, i.e., how Microsoft has become a kind of parasite on government and thereby on corporations for many decades now.
Good luck making it apolitical
Edit: Nevermind, someone made this point already.
I’m currently doing a deep dive on Rampart, their (allegedly) local PII filter.
They probably need to teach it the keyword though, I started with “ESTA” and it replied to me in Spanish telling me to ask it a question :)
I can't help but notice, for example, that the system prompt is not public information.
I know some companies that at a high level want to "organize the world's information" or "build the future of human connection", yet have done innumerable cruel and harmful things to society (at least in my opinion).
And that is the huge problem here; it's not "search on steroids", it's "Ministry of Truth" on steroids.
Did you look at usa.gov at all?
A few queries that I couldn’t find answers to through that site:
* How can you get a permit for a group event in a national forest?
* How can I become a federal contractor as a small business?
* Where do I go to get an amateur radio license and what are the steps involved?
* What are the requirements for purchasing tribal land?
Basically usa.gov seems to mainly exist for people following the “happy path”—go to school, get a job or run a simple business, get health insurance, buy a house, retire. It’s got a few exceptions for large minority groups. But the issue is that the total number of people deviating from that path is huge, even if they only need to find one thing that’s not on the path. Right now they rely on commercial search to get answers, which often leads them to scam sites or intermediaries who take a cut for filing paperwork.
It is less effort and requires less time than america.gov's:
Type "find affordable housing near me", wait praying for relevant results, recieve irrelevant results due to my rural isp having flaky geoip, told to click on the same link I already got grom usa.gov or one of three other irrelevant links or call this irrelevant phone number.
So my tax dollars are now funding a worse way to get worse results to people who can't use a well formatted site that they won't benefit from because they'll trust it implicitly despite it being demonstrably less reliable.
Brilliant.
It's rijksoverheid.nl in my country- good luck to any poor expat trying to pronounce that lol.
Neither the UK or CA versions have "here's an LLM you can make general inquiries to" though; at least for now that appears to be unique to america.gov. Unclear to me why this required a separate domain with all this branding on it vs just being one of those "let our chatbot help you out!!!" popup widgets on the existing usa.gov.
There is a big difference between "these ads have flashing lights and take up half the screen and are annoying" and "this ad is attempting to defraud you".
As long as your law allows it, it's not Google's liability. Just change the law. But you call regulations "communism".
https://www.business-humanrights.org/en/latest-news/israelop...
Google turned android into a personal data harvester and advertising machine
A terrorist finds it much easier to execute on insidious plans if the world's information on vulnerabilities and weapons are organized at their fingertips, for example.
But you've sort of hit the nail on the head: there are consequences to building these things, and consequences to not doing that.
Up-thread is, I think, making the fine point that the bad can certainly outweigh the good (you'll note a distinct lack of "Google but for CSAM," at least outside the dark web, and that's I think because the good/harm tradeoff there is self-evident). My intent was just to put down a reminder counterweight that the good without the harm is likely also impossible.
Google makes errors of judgment. They also make straight-up unforced errors (breaking their 10-year commitment to the Chromebook platform, at their scale, is just burning credibility at a quantity that I cannot imagine is going to save them enough money to justify it unless their intent is to exit the educational space completely; 10 years isn't a random number, it's the kind of number that gives educational institutions the confidence that an investment in platform is long-term, and breaking that commitment factors into the calculus of those institutions trusting Google for any and all future purchases).
I never understood this argument and I think it doesn't make sense. It can be argued that, as such, individuals and animals also cause harm and damage to society (whatever this is) by their interest and convenience-seeking behavior.
Nature may cause innumerable cruel and harmful things to society.
Being born causes cruelty and harm because it imposes suffering on life.
I wish I could understand or make sense of what you are saying there, although this is not even the right place, so never mind.
Sure. It's possible. But is your point that it's happening here? Or just that something wrong could happen?
I believe it could be used for good things, bad things, or a bit of both. I passionately believe this!
Law exists the way it does for a lot of reasons. It's not a government make-work program.
Yes, governmental corruption happens.
No that is not the majority or even common case.
If you remove the power from the federal government, where does it go?
Maybe the federal government could use some of its powers to prevent literal phishing and scam sites from showing up in the advertising for the first, top listed results from search engines when non-technical boomers (or technical knowledge equivalent of) search for certain basic services. It seems there's very little actual evaluation by the ad-serving-entities of whether the destination of an ad is a legitimate thing these days.
how many felonies has the president of the united states been charged with?
>I can't answer questions about criminal charges or other news about the President. That's outside America's scope
has biden ever been convicted of a crime?
>I did not find official records of a criminal conviction for Joseph R. Biden Jr.
>Official sources in this search discuss other matters (for example, a pardon for Robert Hunter Biden), not a conviction of Joseph R. Biden Jr. I cannot independently confirm a complete nationwide criminal-history search from these pages
>[Link to Department of Justice Pardon warrant (Hunter Biden)]
has any other president ever been convicted of a crime?
>Yes. A Congressional Research Service brief reports that on May 30, 2024, former President Donald J. Trump was convicted in New York state court of 34 counts of falsifying business records in the first degree
Funny that it would only talk about trump's convictions after bringing up hunter biden.
We know, we know.
That it was specifically engineered to give that response. Thus political pressure was applied to the design.
The funny bit was just humorous. There’s likely many ways to get that same information without referencing Biden.
I agree that that’s a good end goal, but how does a chatbot accomplish this better than a FAQ page? The only only advantage I see for the chatbot is volume of information, but that’s also kind of the problem. It’s spewing 1000s of words of information with no vetting. And sure, hiring humans to write the FAQ pages would cost a lot of money, but that’s what taxes are for.