11–20 of 235 posts
There should be some warning if some "fact" is only supported by one or very few obscure sources. The strength of the sources should be clearly indicated in the answers to help users gauge how trustworthy the info is.
There should be some warning if some "fact" is only supported by one or very few obscure sources. The strength of the sources should be clearly indicated in the answers to help users gauge how trustworthy the info is.
But you can still just generate any arbitrary amount of information to support the ‘fact’ LLMs are very good at this clearly
It was SOOOOO successful with search, right?
file:///Users/GermaTW1/BBC%20Dropbox/Thomas%20Germain/A%20Downloads%20and%20Documents/2026/And%20there's%20evidence%20that%20AI%20tools%20are%20being%20manipulated%20on%20a%20wide%20scale.
Seems like a lot of entities are "quietly" doing things these days. The llm-ification of every piece of text on the internet is driving me crazy
Earlier quoted context omitted.
But you can still just generate any arbitrary amount of information to support the ‘fact’ LLMs are very good at this clearly
The strength of the sources are not a question of quantity. A hundred obscure blog post have not the same strength as one wikipedia link, because the latter is more trustworthy. There could be some indication beside the info showing the strength of the sources (how many major trustworthy sources support it, etc.).
Yeah, the internet seems like a big poison pill. Training on the whole internet feels like citing the National Enquirer (or the Daily Mail?) for a school essay. Having an archive of "curated" training data seems like it is going to be important. Otherwise you need "AS" (artificial skepticism) introduced into future models. ("But I read it on the internet!", ha ha.) Or perhaps there are ways to bucket training data su…
It's not, though, because the refutations are in the training data too. This isn't actually the problem being described.
The weights in the LLM are fine. It's that the task the LLM is being asked to do is to search and summarize new content that isn't in its training data. And it does it too much like a naive reader and not enough like a cynical HN commenter.
But that's a problem with prompt writing, not training. It's also of a piece with most of the other complaints about current AI solutions, really: AI still lacks the "context" that an experienced human is going to apply, so it doesn't know when it's supposed to reason and when it's supposed to repeat.
If you were to ask it "Is this site correct or is it just spin?" it will probably get it right. But it doesn't know to ask itself that question if it's not in the prompt somewhere.
AI is such garbage. You can't use it for anything.