To give a serious answer to your serious reply to my snarky post: I think we're in an AI hype cycle, I don't deny that AI is sometimes useful but it seems like everyone and their dog is trying to add some kind of "AI" to their product now, without thinking through either the value to the end user, or any privacy implications. They've just heard that AI = $$$ and are running with it. In some cases it even becomes one more way to slurp up everyone's data for ads and training yet more AI.
Even Microsoft Recall could be useful for many people, but it should have been opt-in from the start. Personally I'd also wish that AI modules were only installed when you opt in, so you don't take up valuable space (especially on mobile) with a massive matrix of model weights unless the user actually asks for that.
I am responsible for the Firefox Translation feature (with my dev partner Andre Natal) and I'm proud to have introduced a great "AI" feature. Firefox Translation is powered by small language models (and a transformer architecture) that gives you 100% local translation with a minimal of hassle. It's the opposite of AI-hype and when I was at Mozilla building this, everyone was on the same page as me with using AI only where appropriate and not as a hype mechanism. I left Mozilla about a year ago so perhaps things have changed, but IMO you don't have to worry about Mozilla treating AI as $$$ (mostly because it's actually a huge cost center, I couldn't even get budget to train up beyond about 10 languages when Firefox is available in 100 languages.)
Even Microsoft Recall could be useful for many people, but it should have been opt-in from the start. Personally I'd also wish that AI modules were only installed when you opt in, so you don't take up valuable space (especially on mobile) with a massive matrix of model weights unless the user actually asks for that.