I'm still of the opinion that just swapping current phone assistants with LLMs is naive and bad. For one thing, they need to make sure they don't break current

I'm still of the opinion that just swapping current phone assistants with LLMs is naive and bad. For one thing, they need to make sure they don't break current functionality, like for setting timers and reminders, and and doing math calculations (which LLMs are not good at unless you get it to pass the query off to a dedicated calculator like Wolfram Alpha), etc. But more importantly, when you use something like ChatGPT through the web interface there's a bunch of warning labels all over to remind you "this thing will make up information." For something with such an easily-accessible interface that's baked-in at an OS level, I think a fail state of "sorry, I don't know how to help with that" is better than stating a bunch of false information as fact.