Large language models have their place, but they’re quite limited and no real substitute for human intelligence or judgement at its best.

That said, abusive corporate environments (and they’re pretty much all abusive) turn people into automatons. Execs can thus wave LLMs around and say, “Look, you’re so useless a machine can do your job.”

And even then it does a shitty job. It misses special cases and causes messes. But it’s cheap in the short term, which is all that matters to the boss’s career. Things will go to shit in a few years, but if you’re good at executive social climbing, you are three promotions away from your bad decisions by the time anyone figures out that’s what they were.