I think that previously there were people with ideas, but without the skill to implement them. Whereas now, they perceive themselves to have the skill to implement them, using AI to actually do the work. I think this probably results in a customer we (or anyone else) might actually assertively not want. When things go wrong, they will possess no actual domain knowledge, and we will end up saying one thing, and their AI will say something different. And who/what will they trust?