So, outside of giving your data to government agencies, you're saying the worst thing Google can do is use ML to turn you into a consumer-zombie? While I mostly share your stance on this, that argument just seems completely ineffective (in part to the sense of exceptionalism you mention). Unless there's some poster child case of this that I'm unaware of, I'm not sure why "big scary ML-injected ads" keep getting brought up when there are potentially more valid concerns out there.

Moving on, the remainder of the argument relies on the assumption of an antagonistic state, which if true, means you have a much bigger and immediate problem in the first place. Google or not, China seems to be doing pretty well as a surveillance state so that doesn't seem too convincing either.

So I don't think that opinions will change without a string of high profile incidents that substantiate both of the concerns you mention.