I'll take researchers' gut feelings more seriously once we're out of the ongoing hype cycle. Right now most people are probably overestimating applicability of current advances, and that colours their predictions. Researchers are just as affected by this.

And it's not like this estimate is based on a lot of concrete reasoning. So yeah, I would expect it to fluctuate wildly at an unexpected development in the field initially, then settle down around a slight change. Which is basically the definition of a hype cycle.