@datatitian Hmm... isn't this parallel to email... where eventually when it becomes mature and popular you'll have a few instances running AI regardless (eg. gmail, outlook, etc)?
AI is not inherently evil, there's a fine balance to achieve between utility and privacy.
Like email, the choice is yours, but at some point you'll have to trust your provider/instance, and those with the advantage of AI (whether for advertising or features) will probably end up with more users.