I suspect ai adoption and development work will be driven by jevon's paradox for the forseeable future. It won't be exactly the same work as before, (see here: fixing software flaws found by an AI), but it'll be interesting.