I wasn't able to find anything about Google's Numerics flow. Would you provide a reference?
> Anybody else active in this field?
Yes. Rigetti Computing has also been working in this area. https://medium.com/rigetti/rigetti-partners-with-cdl-to-driv...
Not sure what your issue with the article is. Review papers are among the most cited and important publications in science. There are entire journals dedicated to them, and their impact factor is often an order of magnitude higher than that of original research journals.
 Just to explain how I read it: You draw an incorrect distinction between the authors doing the summarizing and "actual scientists". The review is in fact written by "actual scientists". I consequently interpreted your statement "maybe I don't get it, the scientists all seem very smart", as snarky.
Admittedly maybe getting it out in nature is a bit of an eyeball grabber but.. Still useful.