Don't stick your head in the sand just cause one fad didn't play out.

I’m not, I’m presently underwhelmed by the examples everyone shows.

I’m yet to see actual productivity result from people paying to talk to chatbots to generate boilerplate.

But I tend to shy away from hypers so the LLM craze is passing me by. I have seen uses of AI/ML that helps recognise objects in images which I have seen it do OK at(and it should because it’s the same image just 10m down the road). A human then reviews the outputs. It also spits out highly inaccurate outputs fairly often that the human is necessary even with a feedback loop.