There have been recent articles by financial journalists pointing out that AI is very, very expensive and, since it is not "true" AI, has many of the problems of the existing use of algorithms. Moreover, the easy bits have been done. Any advance from here is going to be harder and more expensive.
Like self-driving cars, AI is an idea that has been around for a long time. Like self-driving cars, it's likely to be a money pit with no real indication of when, if ever, there'll be a return on the investment.
From the tech. companies POV, they are all terrified that someone else is going to crack the problem and reap the benefits. Since they are awash with money, they can chuck it at the problem. They do need to appease their shareholders and they do that with exactly the same hype all new technology gets.
Companies in general avoid making their problems public so articles and reports are likely to be good...especially in these days of journalism by Xitter and press release rather than investigation.
How long this will keep up is another matter. Companies pulled out of developing self-driving cars because it was good money after bad with no payoff on the horizon. Personally, I see "AI" as the new "Smart" : a buzzword attached to things that really deliver little or no advantage to the "old" way of doing it...but which are considerably more expensive for the user and which facilitate the collection of information about users.
