Norway is actually a good example of this – where pro-social regulatory policies (i.e. beneficial not from the perspective of capital, but from the perspective of actual societal conditions) are used to help mitigate some of the BS that capitalism produces.

Regardless… Yeah, it’s a problem with capitalism. It’s a problem that stems from the literal core of the ‘system’: utilizing ‘capital’ to find opportunities for the creation and extraction of ‘surplus’ from labor and its products.

It’s great that regulation is able to reign in, in some cases, the deeply criminal BS that such a system naturally produces… But it seems like a huge overreach to assume this is possible “globally” (as it would need to be for a blanket statement like that to be true).

What they’re getting towards (one thing, anyways) is that “indistinguishable to the model” and “the same” are two very different things.

IIRC, one possibility is that LLMs which learn from one another will make such incremental changes to what’s considered “acceptable” or “normal” language structuring that, over time, more noticeable linguistic changes begin to emerge that go unnoticed by the models.

As it continues, this phenomena creates a “positive feedback loop” in which the gap progressively widens – still undetected, because the quality of training data is going down – to the point where models basically “collapse” in their effectiveness.

So even if their output is indistinguishable now, how the tech is used (I guess?) will determine whether or not a self-destructive LLM echo chamber is produced.