Let's see....
I've always refused to have any of these devices in my house on two counts: Firstly major privacy concerns, secondly; they are really annoying. It's really hard to disable on a phone as well - no single on/off setting... There is one exception where I occasionally use Google Assistant and that is: in my car...
My previous car had Android Auto which was exceedingly disappointing: The primary function I used it for was to play my kids' random song requests whilst driving.
But it would often go like this:
ME: 'Play Baby Shark'
ASSISTANT: 'You need to set up Youtube Music'
ME: 'Ask Spotify to play baby shark'
ASSISTANT: 'Ok, asking Spotify to play Raining Blood by Slayer'
I can't remember which songs we actually involved but this exact scenario arose more than once. At least the kids found it hilarious!
Sometimes the default song would be the wrong one, so you can say 'Ask Spotify to search for the birdie song' but then you need to use the touch screen to chose which version you actually want.
My latest car is built on Android Automotive (AAOS) which is similar but different to Android Auto (Seriously, who passed these product names through Marketing!?!?) and the assistant on that is a step worse.
ME: 'Ask Spotify to play Baby Shark'
ASSISTANT: 'Sorry, I didn't understand that'
ME: 'Play baby shark'
ASSISTANT: 'You need to set up YouTube Music...'
ME: 'Open Spotify'
ASSISTANT: 'Opening Spotify'
But now the only way I can find to voice search in Spotify is to interact with the touch screen to press the search icon and then press the 'Speech to text' button. Given this is all on a touch screen it requires me to divert my attention from the road to achieve this :-(
At one point one of my children really got into a French band called 'Kids United' and would ask me to play things like 'On Ecrit Sur Les Murs' - It's safe to say that Google Assistant cannot handle non-English words when set to English. I'd hazard a guess this means that you're a bit stuffed if you drive your car onto the continent unless you want to go the whole hog and talk to your device in the local language as well.
So if Gemini can handle these tasks then it will be a step forward.
However, the whole voice interface needs to get a whole load better to be genuinely useful - For starters I do not need it to repeat back to me everything I've asked it to do. It would also help if it were less disruptive - i.e. when a song is playing I can't queue up another one without it pausing the one being played. Ideally I'd be able to just say "Play agadoo next" and it would add it to the queue silently and without interrupting Baby Shark.
Likewise I can now fairly reliably press the mic button on my steering wheel and say "navigate to Westward Ho!" but when there are multiple route options and I want to select the one which is only 5 mins slower but 20 miles shorter, if there's a way to do this verbally I haven't figured it out yet so it's back to interacting with the touch screen again (Side rant: I hate touchscreens and they are nothing short of dangerous in cars. Bring back physical buttons!).
So let's see what Gemini brings. If I can control all the bells and whistles of my car by talking to it in plain English then in spite of my general AI/LLM scepticism I might be persuaded to use it,