You say all of that as if it's a good thing.
Objectively, Google is doing just fine as far as I know. So my personal subjective experience is just that: anecdotal.
However, over the last 6 years I have found myself abandoning Google's products en masse, with the most notable being search.
Where Google really shined in its application of ML IMO was eliminating spam. Since then they have tried to use ML to solve all sorts of problems and it has only served to make their products worse.
I switched away from search because, at some point, it started trying to infer some sort of meaning from my keywords when all I want are 1:1 keyword matches ranked by relevance, with exclusion and grouping operators. I started using DuckDuckGo not because of privacy concerns (ok maybe a little bit because of privacy concerns), but because I started to find that it was giving me more relevant results.
I use GMail for work and it's fine, but even that widely popular product doesn't seem any better than the alternatives these days. I still have an ancient Yahoo mail account from the 90s and I use Tutanota for other personal email addresses. Granted, I'm not an email power user so I don't really use tags or other advanced features. Maybe GMail really shines there. But for a simple email client with good spam filtering, GMail is no better than alternatives these days IMO.
Google docs and sheets are OK. I'm not sure how those are leveraging ML but it's the same deal as with GMail. I've used Office365, G-Suite and Libre Office running locally and as for features the only difference in functionality that I can observe is cloud vs local; if that's not a factor then they are all interchangeable.
So I don't see how being "AI First" has improved the end user experience. I liked their products much better 10 years ago.