Your selected quotes do have a survivorship bias. You can find these same kind of predictions for technologies that did fail. NFTs being a recent example of much hype, no follow-through.
The current raft of machine learning applications is broad enough that some things under the AI umbrella will definitely persist (my bets are on image and video generation, and translation being much more useful than chatbots), but time will tell if it really is the next big thing that people like me just don't understand, or 90% hype.