←back to thread

Why language models hallucinate

(openai.com)
277 points simianwords | 8 comments | 06 Sep 25 07:41 UTC | HN request time: 0s | source | bottom
Show context
rhubarbtree ◴[06 Sep 25 21:12 UTC] No.45152883[source]▶
I find this rather oddly phrased.

LLMs hallucinate because they are language models. They are stochastic models of language. They model language, not truth.

If the “truthy” responses are common in their training set for a given prompt, you might be more likely to get something useful as output. Feels like we fell into that idea and said - ok this is useful as an information retrieval tool. And now we use RL to reinforce that useful behaviour. But still, it’s a (biased) language model.

I don’t think that’s how humans work. There’s more to it. We need a model of language, but it’s not sufficient to explain our mental mechanisms. We have other ways of thinking than generating language fragments.

Trying to eliminate cases where a stochastic model the size of an LLM gives “undesirable” or “untrue” responses seems rather odd.

replies(9): >>45152948 #>>45153052 #>>45153156 #>>45153672 #>>45153695 #>>45153785 #>>45154058 #>>45154227 #>>45156698 #
1. utyop22 ◴[06 Sep 25 21:32 UTC] No.45153052[source]▶
The reality is, language itself does not capture the entirety of what is really going on. And I'd get argue its the poorest way of expressing - but one that enables transmission through various mediums efficiently on a cost basis.

E.g. when I explain a concept, what comes to my mind is not a string of letters and words. There is a mix of imagery and even sounds that I may have acquired from learning about a concept - then I translate that into text so it can be communicated.

Theres a reason why people use native subtitles when watching netflix - text complements imagery and sounds.

replies(2): >>45153376 #>>45156472 #
2. pawelmurias ◴[06 Sep 25 22:14 UTC] No.45153376[source]▶
I would assume most people use native subtitles when it's hard to understand what words the actors said.
replies(4): >>45153422 #>>45153462 #>>45153783 #>>45153969 #
3. utyop22 ◴[06 Sep 25 22:21 UTC] No.45153422[source]▶
No that is not the reason.

People watch Netflix to switch their brain off - having the text there helps along with the visual and sound to deliver the content. However, text is inferior to both visual and sound as a delivery mechanism.

replies(1): >>45153651 #
4. jibal ◴[06 Sep 25 22:25 UTC] No.45153462[source]▶
That's why I do.
5. keanebean86 ◴[06 Sep 25 22:53 UTC] No.45153651{3}[source]▶
Subtitles increase the signal to noise ratio. At least in our house. We have to keep the tv low to not wake the child. A volume of 10 with subtitles is similar to volume at 16 without subtitles.
6. ekianjo ◴[06 Sep 25 23:54 UTC] No.45153969[source]▶
Yeah because modern filmmakers make it very hard to hear dialogs for some reason and actors are encouraged to mumble. If I remember correctly even Nolan admitted it.
replies(1): >>45154595 #
7. jibal ◴[07 Sep 25 01:42 UTC] No.45154595{3}[source]▶
And they often speak very quickly--I often rewind to catch critical plot points. It's a lot different from a stage play, where actors enunciate so clearly. (Not that I want stage cadence and booming voices from a film ... they are different art forms.)

Also I watch of English language material that uses accents quite different from what my ears are tuned to.

8. kelnos ◴[07 Sep 25 08:36 UTC] No.45156472[source]▶
I use subtitles becomes sometimes I have trouble understanding the actors. I believe I read something that suggested that the sound mix in movies and cinematic TV shows has changed a lot in the past couple decades, and a result is that it's harder to understand dialogue.

I don't like this; I find my eyes spending more time than I'd like on the text, and not enough on the visual imagery on the rest of the screen. If I truly wanted more text, I'd just read a book.