When the choice is learn to use AI to speed up projects, or to see your department outsourced to other companies who are suddenly flocking like vultures, with shiny AI-written brochures making big promises to do things more cheaply with AI, there really is no choice. I'm within ten years of retirement, very specialised, so I'm just trying to hang on while I can - I can't afford to be a purist and refuse to use it.
It has massive limitations and you have to learn how to use it properly. It forgets earlier instructions when going through a long task. You need to keep refreshing where you are with it and be prepared for the stage where you take over and check everything to be long. Yes you can end up coding things more quickly, but in more like half the time rather than a tenth of the time, because you have to check and test and check and test, over and over again.
It's terrifying to think how much dross there must be out there now, from people accepting AI's first answer and not understanding its limitations.
If you're using it for research, ask it about things you really know about first - that's a good way to really learn just how much it makes up out of nothing. Also, ask it the same thing twice or more in separate sessions, and see how the answers can change.
Remember, when you say "write code that does this" to AI, what it hears is actually "put together some strings of letters and numbers to make something that looks like code". The mimicry can be nearly perfect and really useful, but it's still not properly writing its own code, and it doesn't properly understand it.