Things advertised on television that aren't clearly transactional or a consumer brand are nonsense. (e.g. IBM, GE, "Look King has a new game!", "ADM: Supermarket to the World", ...) It's a very bad smell but the time scale for the corpse to be discovered isn't enough to be good for short selling.
Health care is also a big field for scams. Years ago I ran into somebody who dropped out of my grad school to form a drug discovery startup. He told me he was going to cure cancer. I told him "You know, I'd believe you if you were going to cure Tourette's Syndrome. Everybody and his sister tells me that they are going to cure cancer."
Cancer is not one disease, so there is not going to be one "cure". Progress in cancer treatment is almost always evolutionary rather than revolutionary. Between the fear of death and the money involved, people will be attracted to the domain and it will get press.
Academic A.I. is "mostly nonsense", where the technical definition of "mostly" is >50%. There is significant research, but the fundamental problem is that (almost) everybody is trying to improve the performance of algorithms on the same few training sets, but little work is going on into how to make good training sets quickly.
Commerce has the same problem.
Back in the bad old days, to make an "expert system" you had to compile thousands of rules painstakingly. Now to make a "ML model" you have to compile tens of thousands of training examples specific to your use case. Sure, once you have the data you can run experiments forever ("the race to A.I.") but you don't get to the finish line unless you have the data for your task. In particular you are going to have to split the kind of hairs that make neurotypicals pull their hair out, or it is just going to be "garbage in garbage out".)
Biz folks are highly resistant to making training sets. They say they don't have the money or the time.
Funny, they seem to think that a monthly pass to see unlimited movies that costs less than seeing a movie once is a good idea... Or that a business that offers half-priced taxi rides is a good idea...
But make a training set? That's like extending internet service to people who don't have it or developing a small jet that beats the 737.