Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI False information rate for news nearly doubles in one year

newsguardtech.com

Post reply on HN
Print view View original

61–70 of 85 posts

Re: AI False information rate for news nearly doubles in one year

#61

Post by simonw »

jrflowers wrote: post #25
simonw wrote: post #15

I'm a bit suspicious of this report - they don't reveal nearly enough about their methodology for me to evaluate how credible this is. When it says "The 10 leading AI tools repeated false information on topics in the news more than one third of the time — 35 percent — in August 2025, up from 18 percent in August 2024" - 35% of what ? Their previous 2024 report refused to even distinguish between different tools - mix…

>35% of what? Well it says 35% of the time so I would guess that they’re talking about the number of incidents in a given time frame. For example if you asked me what color the sky is ten times and I said “carrot” four times, you could say that my answer is “carrot” 40% of the time

But what's an "incident"?

There is an enormous difference between "35% of times a user asks a question about news" and "35% of the time against our deliberately hand-picked collection of test questions that we have not published".

Re: AI False information rate for news nearly doubles in one year

#62

Post by simonw »

selcuka wrote: post #32
simonw wrote: post #22

Earlier quoted context omitted.

They seem to still be allowing Google to crawl them, unsurprisingly. Advantage Gemini.

No, they explicitly block Gemini as well: User-agent: Google-Extended Disallow: / Gemini still uses the same user agent, but it has a different robots.txt entry (Google-Extended) [1]: > Google-Extended is a standalone product token that web publishers can use to manage whether content Google crawls from their sites may be used for training future generations of Gemini models that power Gemini Apps and Vertex AI API f…

Honestly I feel like "training" is a bit of a distraction at this point. For a lot of types of content RAG-style search is much more important.

I imagine many of the orgs that are blocking "training" don't understand the difference between training and inference-time tool-based context extension (which really needs an agreed upon name, it's hard to talk about right now).

Re: AI False information rate for news nearly doubles in one year

#63

Post by furyofantares »

simianwords wrote: post #56
ehnto wrote: post #53

Earlier quoted context omitted.

How would you know, generally speaking? Factuality online is always subjective, based on a thing someone said, or that you observed, and you are putting trust in the source. Whether you googled it, or AI googled it and generated an explanation, you are trusting the source. You can influence it so easily with your inputs as well. You could easily, accidentally point it's search toward the searches someone is more like…

These are good caveats and questions worth asking. But it doesn't take away my main point - grok is useful in countering false claims and it is accurate very frequently.

"distinguishing facts from falsehoods" is so broad to encompass all of scientific and philosophical endeavor. I know they mean something much more narrow, but "fact checking" is also fairly broad.

You probably mean things where the truth is widely known and you should have no trouble sorting it out if you put a little effort in. TFA has nearly zero information about what sorts of things they're talking about here, and clicking around the website I found only a _little_ more info.

Re: AI False information rate for news nearly doubles in one year

#64

Post by tomhow »

pgbuttplug wrote: post #4
baby wrote: post #2

And now that Elon is in charge of "fixing" grok every time it's too "woke" this is probably going to get worse: https://x.com/kvistp/status/1967115650276749439

[flagged]

We've banned this account. Please don't use troll accounts for engaging in ideological battle. People who do this eventually have their main account banned too.

Re: AI False information rate for news nearly doubles in one year

#65

Post by snailmailman »

simianwords wrote: post #47
snailmailman wrote: post #3

This isn’t surprising to me at all. These services put too much trust in “the most common answer” when that might not be the correct answer. Just because people think one thing doesn’t make it true. It’s super easy to spread misinformation online. And if you can SEO to the top, the AI will think your site is correct. I see factually incorrect “ai summaries” in search results all the time and see that it cites ai-gene…

>This isn’t surprising to me at all. These services put too much trust in “the most common answer” when that might not be the correct answer. Just because people think one thing doesn’t make it true. It’s super easy to spread misinformation online. And if you can SEO to the top, the AI will think your site is correct. Yeah I want the answer that the world has converged on and not some looney answer. It seems like you…

I have used various AI models. They often aren't good at fact checking. They hallucinate mistakes all the time and often even insist that they are correct when they are not. Some models are better at this, but the ones used in "ai summaries" on various search engines aren't in my experience.

I want the answer that is actually correct. I dont want the ai generated "answer" that has the most SEO, or has appeared often in results, but was entirely made up. When i say "the most common answer" can be incorrect - there are often recent topics where reliable sources don't exist yet. As an example, I've been playing through silksong recently, and had some questions about some late-game content. The official wiki is extremely incomplete, as the game is brand new. I had several questions, and asked a few AI's and search engines. I got completely wrong information a few days ago, but trying the the questions again today, some AI's are only now giving me the right answer. But the correct information simply doesn't appear online enough or in the right places.

(To avoid game spoilers - I wanted to know the quantity of an item, because i wanted to know if i found them all. the answer is "at least 4" because i found 4 of this item while playing. Nearly every AI said "3" a few days ago, and many still do.)

Re: AI False information rate for news nearly doubles in one year

#66

Post by simianwords »

ehnto wrote: post #53
simianwords wrote: post #43

Earlier quoted context omitted.

I routinely use AI to fact check claims and it works extremely well.

How would you know, generally speaking? Factuality online is always subjective, based on a thing someone said, or that you observed, and you are putting trust in the source. Whether you googled it, or AI googled it and generated an explanation, you are trusting the source. You can influence it so easily with your inputs as well. You could easily, accidentally point it's search toward the searches someone is more like…

To support my claims, here is a typical example: https://x.com/vlastimilve/status/1967187345641558272?s=46

I didn’t cherry pick it because obviously it’s not that impressive. But it works - it is professional and gives as much information as it can be confident giving.

You said you can influence it etc but clearly it is hard. Sure it is not perfect but I find that it helps reducing fake news if anything.

You are right that for things that are more subtle you can lead it to certain answers. But we must acknowledge that debates that are not fully solved and have subtleties are not exactly relevant to “spreading misinformation”.

Re: AI False information rate for news nearly doubles in one year

#67

Post by ehnto »

simianwords wrote: post #66
ehnto wrote: post #53

Earlier quoted context omitted.

How would you know, generally speaking? Factuality online is always subjective, based on a thing someone said, or that you observed, and you are putting trust in the source. Whether you googled it, or AI googled it and generated an explanation, you are trusting the source. You can influence it so easily with your inputs as well. You could easily, accidentally point it's search toward the searches someone is more like…

To support my claims, here is a typical example: https://x.com/vlastimilve/status/1967187345641558272?s=46 I didn’t cherry pick it because obviously it’s not that impressive. But it works - it is professional and gives as much information as it can be confident giving. You said you can influence it etc but clearly it is hard. Sure it is not perfect but I find that it helps reducing fake news if anything. You are righ…

It's possible I'm blocking the AI widget as I don't see it, is it Grok replying with a fact check of the claim?

I think that is good in general, I do, I am just weary that it is still shakey ground. The company behind Grok has clear incentives to influence the output of those fact checks as one example. I am also weary of it always eventually conforming to the vox populi as it ingests the internet firehouse. Leading to the loudest, most prolific voices thus views always being more represented.

Re: AI False information rate for news nearly doubles in one year

#68

Post by add-sub-mul-div »

simianwords wrote: post #44
baby wrote: post #2

And now that Elon is in charge of "fixing" grok every time it's too "woke" this is probably going to get worse: https://x.com/kvistp/status/1967115650276749439

Shouldn't this tell you that AI is at its core factually correct than not correct because it is much harder to hack it?

Musk and his crew failing at something isn't evidence that the task was hard.

Re: AI False information rate for news nearly doubles in one year

#69

Post by selcuka »

simonw wrote: post #62
selcuka wrote: post #32

Earlier quoted context omitted.

No, they explicitly block Gemini as well: User-agent: Google-Extended Disallow: / Gemini still uses the same user agent, but it has a different robots.txt entry (Google-Extended) [1]: > Google-Extended is a standalone product token that web publishers can use to manage whether content Google crawls from their sites may be used for training future generations of Gemini models that power Gemini Apps and Vertex AI API f…

Honestly I feel like "training" is a bit of a distraction at this point. For a lot of types of content RAG-style search is much more important. I imagine many of the orgs that are blocking "training" don't understand the difference between training and inference-time tool-based context extension (which really needs an agreed upon name, it's hard to talk about right now).

My understanding is that it also affects RAG ("grounding" in Google terminology):

> [...] and for grounding (providing content from the Google Search index to the model at prompt time to improve factuality and relevancy) in Gemini Apps and Grounding with Google Search on Vertex AI.

So they seem to be blocking both training and RAG while still allowing search engine indexing.

Re: AI False information rate for news nearly doubles in one year

#70

Post by simianwords »

add-sub-mul-div wrote: post #68
simianwords wrote: post #44

Earlier quoted context omitted.

Shouldn't this tell you that AI is at its core factually correct than not correct because it is much harder to hack it?

Musk and his crew failing at something isn't evidence that the task was hard.

Why not?
Post reply on HN