OK, there's a lot of negative comments here, but this is a great idea at a high level. It really is hard to figure out where to do a thing and it's also very easy to get phished. If they can figure out how to help people get all the services they're eligible for that'll be an amazing improvement for the people.
I agree in general, but how do you resolve the problem that people will take the output of the chatbot as authoritative even when it could possibly have errors or miss important nuance? People being lead astray with information from the government could lead to very serious consequences, up to and including imprisonment.
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.
> but how do you resolve the problem that people will take the output of the chatbot as authoritative even when it could possibly have errors or miss important nuance
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.
Giving people a big list of life events isn’t nearly as easy as allowing them to just type in what’s going on and having an LLM figure out what’s relevant to them.
It isn’t nearly as easy but it’s more reliable whereas using an llm is prone to unknown hallucinations (also ignoring the obvious ethical considerations of llms).
15 years ago I was talking to my HR rep about covering my child on healthcare. Her other parent and I were not married. “No you can’t cover her!” “Why not, I’m one of the two parents?” “Because you’re not married!” “So what?”
Cool, but the information on government websites generally meets a higher standard than just ‘making shit up all the time’ so it’s not really a relevant comparison.
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.
Yes, but when a human makes shit up, they're still legally liable for what they said, and depending on context can be compelled to fix the issue or lose their job.
If an LLM makes shit up, your one and only recourse is "computer says no".
We were discussing zoning in another thread, I think your comment here is very appropriate. People make up all sorts of reasons to oppose land use, which is part of the reason municipal land control is so dangerous.
I encounter it daily in the LLM outputs my coworkers produce.
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?
My coworkers can be wrong, sure. But no, they weren't making stuff up pre-LLMs. But now making stuff up is automated and easy, so we've gone from simply mistaken to piles of "what the hell is this even trying to say?"
Maybe some things just are not easy. Maybe they should not be. Maybe you should be able to have basic citizenship skills to survive in a modern democ… Erm, whatever it is you guys are doing now.
As far as I can tell, the america.gov LLM is mostly providing links to the landing pages you would have found through classical web search. Perhaps with a little bit of clarifying information when there are multiple choices.
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!
maherbeg · · focus · HN ↗
gthrow12345 · · focus · HN ↗
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.
rayiner · · focus · HN ↗
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.
LelouBil · · focus · HN ↗
I France service-public.gouv.fr is amazing
<a href="https://www.service-public.gouv.fr/particuliers/vosdroits/F16225" rel="nofollow">https://www.service-public.gouv.fr/particuliers/vosdroits/F1...
rayiner · · focus · HN ↗
doc_ick · · focus · HN ↗
irishcoffee · · focus · HN ↗
Humans also make shit up all the time.
astafrig · · focus · HN ↗
irishcoffee · · focus · HN ↗
lkbm · · focus · HN ↗
5upplied_demand · · focus · HN ↗
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.
dansquizsoft · · focus · HN ↗
unrented7977 · · focus · HN ↗
If an LLM makes shit up, your one and only recourse is "computer says no".
Schiendelman · · focus · HN ↗
rayiner · · focus · HN ↗
kijashdkujdfhas · · focus · HN ↗
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?
rerdavies · · focus · HN ↗
kijashdkujdfhas · · focus · HN ↗
delis-thumbs-7e · · focus · HN ↗
drstewart · · focus · HN ↗
Okay. So why should government websites like the French one be well designed? Maybe it should not be.
LelouBil · · focus · HN ↗
rerdavies · · focus · HN ↗
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!