I think that's relatively emergent too though! BERT never really did that (at least to my recollection), presumably because its training was never sufficient for it to develop corrective reasoning in a chain of thought.
In a general sense I mean something like "arising through complex interaction between parts", like the murmurations of certain birds or unexpected mechanics found in physics sandbox games.
In this specific instance I mean I don't think it was obvious given the structure of the model, it only "emerged" when we applied massive scale, which allowed for unexpected interactions to happen in the deep networks involved.
I don't know what the parent poster means, but if you look at how LLMs are trained, and how it's trained on human communication, you should find it interesting that humans often correct themselves. They also often create facts out of nowhere, and do this crazy thing called dreaming.
If you're an LLM trying to symbolically emulate human behavior, especially when coding, you're going to see git commits, where people write a bunch of code and then go, oh wait, I should fix that!
From the perspective of an outside observer, making mistakes, and then correcting the mistakes, is how you code.
In fact, an outside observer might think that this is how you teach people to code. You make a mistake, then you say, oh, here's the mistake I made, and here's why, and then you fix it.
So I think true emergent behavior will be demonstrated, when an LLM says "Hey wait a minute! They're actually making real mistakes, and they're actually correcting, it's them not on purpose?!", and then stops hallucinating, and stops making mistakes as it codes.
The same is true of bug trackers. You submit a bug request and you often have a patch. Then people discuss it. From a MLM's perspective, you're supposed to write poor code, or code that's not perfect, and then have a conversation.
If I had the resources to train a large-scale LLM, I would clone, for example, GitHub. I would then remove everything where people are fixing broken code.
The outcome would be very interesting.
I've been looking at ways to make enhanced long-term memory stores for LLMs, and there's lots of problems with shifting symbolic relationships if you do it wrong, but definitely once there's true long-term memory, and adaptive behavio,r I think that's the only way you're going to get true real emergent behavior.
In general, I completely agree with you. Personally, as a human: I have fallible memory, beliefs that are not as sound as I may believe, and all of those notwithstanding, I may not correctly reason, even if my "inputs" were perfect. Let's not even get into cognitive biases, etc. I am acutely aware of my, and other humans' fallibilities.
I think when people are irritated by the "hallucination" aspect of LLMs, they are often running into something of a slightly different nature. I mean, firstly, there seems to be a higher-than-normal ratio of these "little mistakes". And secondly, there are some pretty odd ones - e.g. in my team, Claude regularly just straight up makes up Jira ticket numbers, and then refers to them with high confidence. I guess what I'm saying is that a human would probably not just make up an id, and run with it (they may be off by one, or mix two up, etc). In my opinion, these can be successfully treated, but I guess you can never fully eliminate the tendency that irks people.
I mean sure, but BERT can be and often is used as a next token predictor / generator.
I could have used any series of NLP examples, the point is this CoT behavior only emerges when you get to a certain scale (and training style, presumably).
