Post #2087511

2026-05-06 14:48 UTC

The problem is less that the system would somehow ignore that part of the prompt and more that “hallucinate” or “make stuff up” aren’t special subroutines that get called on demand when prompted by an idiot, they’re descriptive of what an LLM does all the time. It’s following statistical patterns in a matrix created by the training data and reinforcement processes. Theoretically if the people responsible for that training and reinforcement did their jobs well then those patterns should only include true statements but if it was that easy then you wouldn’t have [insert the entire intellectual history of the human species]. Even if you assume that the AI boosters are completely right and that the LLM inference process is directly analogous to how people think, does saying “don’t fuck up” actually make people less likely to fuck up? Like, the kind of errors you’re looking at here aren’t generated by some separate process. Someone who misremembers a fact doesn’t know they’ve misremembered until they get called out on the error either by someone else with a better memory or reality imposing the consequence of being wrong. Similarly the LLM isn’t doing anything special when it spits out bullshit.

Replies (2)

@Architeuthis@awful.systems 2026-05-07 04:54

Theoretically if the people responsible for that training and reinforcement did their jobs well then those patterns should only include true statements That would only work if inference was some sort of massive if-the-else process. Hallucinations are downstream of neural networks’ ability to generalize from the dataset examples, they aren’t going anywhere.

Open ##2131235
@scruiser@awful.systems 2026-05-09 16:57

Theoretically if the people responsible for that training and reinforcement did their jobs well then those patterns should only include true statements but if it was that easy then you wouldn’t have [insert the entire intellectual history of the human species]. I’m chiming in to agree with Architeuthis and mention a citation explaining more. LLMs have a hard minimum rate of hallucinations based on the rate of “monofacts” in their training data (arxiv.org/html/2502.08666v1). Basically, facts that appear independently and only once in the training data cause the LLM to “learn” that you can have a certain rate of disconnected “facts” that appear nowhere else, and cause it to in turn generate output similar to that, which in practice is basically random and thus basically guaranteed to be false. And as Architeuthis says, the ability of LLMs to “generalize” basically means they compose true information together in ways that is sometimes false. So to the extent you want your LLM to ever “generalize”, you also get an unavoidable minimum of hallucinations that way. So yeah, even given an even more absurdly big training data source that was also magically perfectly curated you wouldn’t be able to iron out the intrinsic flaws of LLMs.

Open ##2293545