Strong take and honestly, I’ve seen both sides.
In one project, AI looked impressive in demos but didn’t move any real user metrics because the core workflow was still clunky. We were optimizing the “intelligence” instead of fixing the user journey.
In another case, we stripped things back, defined a very clear user pain point first, and then added AI in a narrow, focused way. That’s where it actually improved engagement and saved time.
So yeah AI didn’t fix bad product thinking. It amplified it. When the foundation was solid, AI felt like magic. When it wasn’t, it just added noise.