We go through this hype cycle every time a technology grows exponentially. Every salesperson shouts "It's going to the moon!" and it goes to Radio Shack and Walmart instead.
It actually turns out that everything that looks exponential, actually follows a logistic growth curve, constrained by other factors.
We're usually pretty bad at guessing which other factors will kick in to stop the exponential growth, and when. And sometimes the exponential growth kills us anyway, before another factor kicks in.
But the current AI hype will also be governed by other factors. Even if we agree to have complete 100% faith in AI's abilities growing as much as possible, it will hit a hard limit of tasks that it will remain stubbornly bad at.
Where ever that limit currently is, is where the interesting work will happen. Currently it's at "go for more than a few minutes without lying or hallucinating". Someday soon it will be around AI driving on a busy day without smearing an unacceptable number of pedestrians into mush.
AI becoming better will be a long, often sad, always interesting journey.
But whoever is telling you that AI has already replaced you is just trying to keep you off balance, hoping you won't find and connect with your local union representative.