[stavrianos]

I keep seeing people talking about chatgpt hallucinating when it's wrong, but not when it's right. Maybe I've misunderstood, but isn't it just always hallucinating? It's not like the failure-mode is meaningfully different from the successes, except insofar as whether we agree with it, right?

[ggm]

Hallucinating is no different to believe or know or infer: its the wrong label to use for what it does.

Yes, colloquially what it does is hallucinate all the time, and sometimes it lucid. But more factually no, it doesn't hallucinate because there is no "it" there, it's not conscious and you need to have a brain, to hallucinate.

There's no "there" there.

That is the whole of my point: we're using the wrong labels to describe what is happening.

When it comes to explaining and describing "it's like" is one of the WORST ways to go. explanation by analogy or metaphor is a trap. "atoms are like billiard balls BZZZT next" "cells are little bags of water BZZZT next" "panadol 'kills' the pain BZZT no, it doesn't kill anything next"

[pulvinar]

At what point would it be ok to use these words? Will there be some threshold we'll all recognize? Or maybe a bill of rights to force the issue?

I predict acceptance will go the way the ether disappeared-- advancing one funeral at a time.

[starfallg]

'Hallucination' is the term LLM researchers use, and is arguably inaccurate based on the dictionary definition. The actual term for this for behaviour in people is 'confabulation', which is a lot more accurate.

[ggm]

Oh well.. if the LLM researchers are going to coin terms of art, well and good but I dislike this intensely because it invites belief it implies AGI and brain when in fact, its synthesis of new state from a model.

This coining terms of art thing isn't uncommon. Think "brutalist architecture" and remind yourself its "en brute" == raw from the french. It has nothing to do with how "brutal" people think concrete is.

[SanderNL]

Right. We are the ones applying the label after the fact.

[aezart]

I agree. A stopped clock is right twice a day, and that's equally true with these LLMs.