Aside from the aforementioned local models path though, this whole productivity angle (which the above poster loves to shit on btw) also serves to retain jobs. Current data suggests that rather than letting people go, companies are banking on extracting more productivity out of workers, partly because the models are admittedly way overhyped, partly because it's the sane other option to mass layoffs, and partly since these models still need and strongly benefit from in-context steering. And they forever will: the human experience is human by definition, we're the "oracles" to it. How much that will continue to justify employments is still out there though, of course. I do expect a crunch phase, provided there was any actual productivity gain realized to begin with, which in itself is very loosely supported if at all.
Regardless, I don't see the point in not using these, or lying about how good they are, or willfully hating on them. Never helped anyone. Early and quality information however, very much so. If I know the time has come or is actually coming, I can take action accordingly. If I listen to every random social media thread I come across instead, not so much. According to social media, software engineering has been over for 3 years now already. The wolf was not only cried, but turned into a whole musical outright. The extremely dissonant clash of the sentiments "LLMs are pure shit, actually" and "it's like, literally taking our jobs" is not lost on me either.