I think these studies aren't meaningless at all, but the fact that "AI" is a loosely used term means that many people might view even more simple ML methods with skepticism, as opposed to just, say, chat-like LLM tools.
I'm still waiting for the last ML movement to revolutionize business intelligence. Back when regression models were going to give us all forecasting. Turns out garbage in still equals garbage out and there still aren't any silver bullets. The organizations that couldn't get their act together to collect good data about their businesses for traditional analysis methods to work are shock-faced that model-overfitting writ large isn't saving them from their doofus C-suites.
You can't regress to the mean and call it creation. LLMs don't make novel content. This is why all the people using AI-summarizers to understand their boss's AI-expanded micromanaging emails aren't getting anything new done. Anti-compression is going to accelerate climate change.