←back to thread

Ask HN: Any insider takes on Yann LeCun's push against current architectures?

385 points vessenes | 8 comments | 10 Mar 25 19:41 UTC | HN request time: 0.791s | source | bottom

So, Lecun has been quite public saying that he believes LLMs will never fix hallucinations because, essentially, the token choice method at each step leads to runaway errors -- these can't be damped mathematically.

In exchange, he offers the idea that we should have something that is an 'energy minimization' architecture; as I understand it, this would have a concept of the 'energy' of an entire response, and training would try and minimize that.

Which is to say, I don't fully understand this. That said, I'm curious to hear what ML researchers think about Lecun's take, and if there's any engineering done around it. I can't find much after the release of ijepa from his group.

1. TrainedMonkey ◴[14 Mar 25 18:03 UTC] No.43365343[source]▶
This is a somewhat nihilistic take with an optimistic ending. I believe humans will never fix hallucinations. Amount of totally or partially untrue statements people make is significant. Especially in tech, it's rare for people to admit that they do not know something. And yet, despite all of that the progress keeps marching forward and maybe even accelerating.
replies(5): >>43365376 #>>43366746 #>>43367326 #>>43368390 #>>43368923 #
2. ketzo ◴[14 Mar 25 18:06 UTC] No.43365376[source]▶
Yeah, I think a lot of people talk about "fixing hallucinations" as the end goal, rather than "LLMs providing value", which misses the forest for the trees; it's obviously already true that we don't need totally hallucination-free output to get value from these models.
replies(1): >>43368890 #
3. dtnewman ◴[14 Mar 25 20:12 UTC] No.43366746[source]▶
I’m not sure I follow. Sure, people lie, and make stuff up all the time. If an LLM goes and parrots that, then I would argue that it isn’t hallucinating. Hallucinating would be where it makes something up that is not in its training site nor logically deducible from it.
4. esafak ◴[14 Mar 25 21:08 UTC] No.43367326[source]▶
I think most humans are perfectly capable of admitting to themselves when they do not know something. Computers ought to do better.
replies(1): >>43367727 #
5. danielmarkbruce ◴[14 Mar 25 22:00 UTC] No.43367727[source]▶
You must interact with a very different set of humans than most.
6. danielmarkbruce ◴[14 Mar 25 23:25 UTC] No.43368390[source]▶
Once one starts thinking of them as "concept models" rather than language models or fact models, "hallucinations" become something not to be so fixated on. We transform tokens into 12k+ length embeddings... right at the start. They stop being language immediately.

They aren't fact machines. They are concept machines.

7. mdp2021 ◴[15 Mar 25 00:45 UTC] No.43368890[source]▶
Even as language models can partially solve a few problems, we remain with the problem of achieving Artificial General Intelligence, that the presence of LLMs has exacerbated because they so often reveal to be artificial morons.

Intelligence finds solutions - actual, solid solutions.

More than "fixing" hallucinations, the problem is going beyond them (arriving to "sobriety").

8. mdp2021 ◴[15 Mar 25 00:51 UTC] No.43368923[source]▶
Not an argument. "Many people are delirious, yet some people create progress". What is that supposed to imply?