And at the very least, it is much harder for Google to correlate a competitors' account systems and user data with their own profiles they use for ad targeting.
And of course, most importantly: Apple collects a lot less data when you use their apps. For instance, Maps data is never associated with an Apple ID, and significant work is done to fuzz and scramble the anonymous data they store to avoid it being possible to turn it into data about a real person. In comparison, Google Maps logs your location to your Google account every five minutes.
https://www.google.com/amp/s/www.theverge.com/platform/amp/2...
Edit: I can reply anymore but just supply a link of data handling discrimination between paying customers because that's what you imply by "there is not difference between GSuite customers and normal gmail".
It's bullshit
Many of Google's uses of data are not specific to whether or not someone is using G Suite or consumer Gmail as far as I am aware.
https://gsuite.google.com/learn-more/security/security-white...
I definitely don't want my email used to train Google's AI models, which might unpredictably cough up sensitive data since machine learning is pretty much a black box.
A big issue is that at any time, Google may introduce a new "feature" and decide to use your data to power it. I also think of how purchase data gets extracted from your receipts to store in your Google account profile and the like (https://myaccount.google.com/purchases although the text on this page has changed... did they stop?) and I feel a privacy-first service would need to at minimum have switches to control what kind of data processing is performed on their email, and default new features to be off.
What sensitive information are you concerned about coughing up? While I haven't actually looked closely at how smart reply works, it isn't just a simple lstm on text, it tracks contextual info like recipient names, so "hey " on the first line will autocomple to "Hey Todd and Amanda" if you're sending am email to two people named Todd and Amanda.
So there's some tokenization and preprocessing happening, which would avoid most of the potential privacy issues I can think up with something like that (which are all of the form of automoplete linking your name to a private fact).
But it has nothing to do with what the actual policy on customer data is. My guess is that you'd be pleasantly surprised be the policy. I just can't find it documented publicly anywhere, so I don't know that I can share the actual policies. There are people who's job it is to share such things, they're like account managers and support people and they have guidance on the way the things they can say. My guidance is more like "don't unless it's clearly public".
Put another way, the training on data handling is longer than the training on corporate communication. So if I'm this careful with my communication, imagine the ridiculous lengths I go to with user data safety.