> the simple 'perception of having taken the wrong path' by researchers or companies
IMO, neither the plateau nor the perception of "a wrong path" are real. There are too many paths and we have too few humans with adequately capable brains.
Companies talk for the agenda's sake and thus the kick of the surprise. It's a marketing thing.
AI R&D is basically thinking out loud nowadays. It's just the pace of the news.
I believe that most AI development has reached "the end" of a logarithmic curve. The assigned humans will catch up. Then we'll see faster growth again. It takes time to get from one edge to the other or walk along it or explore the area.
The progress is there but it's infinitely small compared to the past years where it was relatively simple to get better results over and over and nobody will get it except if they are sensitized to it.
What kind of major leap in performance do you expect? What do others expect? Be specific and people will tell you whether there is a plateau or not enough hands on deck working on specific problems.