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

Differences in link hallucination and source comprehension across different LLM

mikecaulfield.substack.com

Post reply on HN
Print view View original

21–30 of 47 posts

Re: Differences in link hallucination and source comprehension across different LLM

#21

Post by krzat »

motorest wrote: post #4

Taken from the blog: > Why are we talking about “graduate and PhD-level intelligence” in these systems if they can’t find and verify relevant links — even directly after a search? This is my pet peeves, and recently OpenAI's models seem to have become very militant in how they stand by and push their obviously hallucinated sources. I'm talking about hallucinating answers, when pressed to cite sources they also halluc…

The approach of generating something and then looking for hallucinations is just stupid. To validate the output I have to be an expert. How do I become an expert if rely on LLMs? It's a dead end.

Re: Differences in link hallucination and source comprehension across different LLM

#22

Post by nkrisc »

motorest wrote: post #4

Taken from the blog: > Why are we talking about “graduate and PhD-level intelligence” in these systems if they can’t find and verify relevant links — even directly after a search? This is my pet peeves, and recently OpenAI's models seem to have become very militant in how they stand by and push their obviously hallucinated sources. I'm talking about hallucinating answers, when pressed to cite sources they also halluc…

Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens.

Re: Differences in link hallucination and source comprehension across different LLM

#23

Post by motorest »

krzat wrote: post #21
motorest wrote: post #4

Taken from the blog: > Why are we talking about “graduate and PhD-level intelligence” in these systems if they can’t find and verify relevant links — even directly after a search? This is my pet peeves, and recently OpenAI's models seem to have become very militant in how they stand by and push their obviously hallucinated sources. I'm talking about hallucinating answers, when pressed to cite sources they also halluc…

The approach of generating something and then looking for hallucinations is just stupid. To validate the output I have to be an expert. How do I become an expert if rely on LLMs? It's a dead end.

> The approach of generating something and then looking for hallucinations is just stupid. To validate the output I have to be an expert.

No. You only need to check for sources, and then verify these sources exist and they support the claims.

It's the very definition of "fact".

In some cases, all you need to do is check if a URL that was cited does exist.

Re: Differences in link hallucination and source comprehension across different LLM

#24

Post by motorest »

nkrisc wrote: post #22
motorest wrote: post #4

Taken from the blog: > Why are we talking about “graduate and PhD-level intelligence” in these systems if they can’t find and verify relevant links — even directly after a search? This is my pet peeves, and recently OpenAI's models seem to have become very militant in how they stand by and push their obviously hallucinated sources. I'm talking about hallucinating answers, when pressed to cite sources they also halluc…

Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens.

> Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens.

No. If you prompt it to get a response and then you ask it to cite sources, if it outputs broken links that never existed then it clearly failed to deliver correct output.

Re: Differences in link hallucination and source comprehension across different LLM

#25

Post by simonw »

The key thing I got from this article is that the o3 and Claude 4 projects (I'm differentiating from the models here because the harness of tools around them is critical too) are massively ahead of GPT 4.1 and Gemini 2.5 when it comes to fact checking in a way that benefits from search and web usage.

The o3 finding matches my own experience: https://simonwillison.net/2025/Apr/21/ai-assisted-search/#o3...

Both o3 and Claude 4 have a crucial new ability: they can run tools such as their search tool as part of their "reasoning" phase. I genuinely think this is one of the most exciting new advances in LLMs in the last six months.

Re: Differences in link hallucination and source comprehension across different LLM

#26

Post by nkrisc »

motorest wrote: post #24
nkrisc wrote: post #22

Earlier quoted context omitted.

Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens.

> Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens. No. If you prompt it to get a response and then you ask it to cite sources, if it outputs broken links that never existed then it clearly failed to deliver correct output.

But are the links plausible text given the training data?

If the purpose is to accurately cite sources, how is it even possible to hallucinate them? Seems like folks are expecting way too much from these tools. They are not intelligent. Useful, perhaps.

Re: Differences in link hallucination and source comprehension across different LLM

#27

Post by alkonaut »

vanschelven wrote: post #16
motorest wrote: post #4

Taken from the blog: > Why are we talking about “graduate and PhD-level intelligence” in these systems if they can’t find and verify relevant links — even directly after a search? This is my pet peeves, and recently OpenAI's models seem to have become very militant in how they stand by and push their obviously hallucinated sources. I'm talking about hallucinating answers, when pressed to cite sources they also halluc…

Including literal 404s... As an outsider it has always struck me as absurd that they don't just do the equivalent of wget over all provided sources.

Or why the LLM doesn’t do a lookup into a subset of the training data as a database and reject the output if it seems to be wrong. A billion of the most urls and the entirety of Wikipedia, arkiv and stackoverflow would go a long way.

Re: Differences in link hallucination and source comprehension across different LLM

#28

Post by diego_moita »

I have a strange feeling: it seems that original insights and hallucinations are related. One seems to come very frequently with the other.

I've noticed that o3 is the one that lies with the most conviction (compared to Gemini Pro and Claude Sonnet). It will be the hardest to convince that it is wrong, will invent excuses and complex explanations for its lies, almost to a Trump level of lying and deception.

But it is also the one that provides the most interesting insights, that will look at what others don't see.

There might some kind deep truth in this correlation. Or it might be myself having an hallucination...

Re: Differences in link hallucination and source comprehension across different LLM

#29

Post by piva00 »

genewitch wrote: post #11
Orygin wrote: post #10

Earlier quoted context omitted.

Languages evolve and words get new meanings all the time.

Yeah the meaning was what I said until about 2 weeks ago when someone went viral talking about air travel without a cellphone. Do vulgarities often become accepted?

Some of them do, yes. Tone policing has always been around as well, just surprises me to happen in an online forum about "vulgarities".

Re: Differences in link hallucination and source comprehension across different LLM

#30

Post by Scarblac »

motorest wrote: post #24
nkrisc wrote: post #22

Earlier quoted context omitted.

Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens.

> Seems like the LLM is giving correct output if it’s generating a plausible string of tokens in response to your string of tokens. No. If you prompt it to get a response and then you ask it to cite sources, if it outputs broken links that never existed then it clearly failed to deliver correct output.

Seems that's just expecting things that LLMs were not designed for.

It's a token producer based on trained weights, it doesn't use any sources.

Even if it were "fixed" so that it only generates URLs that exist, it's still incorrect because it did not use any sources so those URLs are not sources.

Post reply on HN