Hey, that’s nice! you’re actually trying to engage, although you’re immediately undermining it with the whole “best scenario” thing where you try to cast my example as somehow doubtful instead of “something we regularly test”. But to answer your question, data management has proven pretty resilient to AI - we do use it, it’s a great way to fill out boilerplate code, but AI currently fundamentally lacks the ability to manage large complex architecture, most adhoc reporting or even middling novel applications (and holy shit it’s bad at dynamic SQL). From your description of your job you’re not doing anything (and I swear this isn’t a value judgement on you personally) particularly interesting - IDK if you’re devops or webdev or appdev or what - but there are segments of software development that have been largely “solved problems” for many years; the humans mostly manage and direct tools, and are kept around to solve the interesting and challenging problems. Within those segments, AI has been extremely effective at integrating into, and with some caveats about how much it screws up in devops accelerating, workflows, in an ideal world freeing up the humans to focus even more of their time on the interesting problems (in practice this has led people to think they can lay off their devops team, with disaster not far behind). In the rest, AI truly is not at the point it can live up to it’s promises, and the cost in tokens trying to integrate it is often astronomical. From how you approach this, I get the impression that you don’t have much experience outside your segment or role - it’s worth not being quite so pig-headed about this when Lemmy, which hosts a population of users heavy with non-mainstream disciplines, shares their own anecdotal experience about this subject. It’s not nearly as cut-and-dry as the massive propaganda engines would have you believe.