I’m not necessarily confrontational about AI adoption. I’m more interested in situations where leadership basically sets you to fail: unrealistic velocity expectation, too much context switching, code sprawl, PR floods, and so on.
My approach was usually to start looking - not frantically, but with some sustained effort. If they persist, then it's increasingly going to be a bad place to work, and I can usually find a better place.
AI does increase productivity if you use it as a correct tool when needed. Measure the productivity gain/loss over the time and refine approach iteratively.