I have a similar experience.

Having used Google home assistant since it came out for all the things that it's good for, and watched its quality fluctuate, I find I really have to be more careful nowadays when I ask it questions because it can go overboard more easily.

Is there a term in AI research where it underappreciates the specificity that is being asked for?

Perhaps the AI could be default prompted, "you are a kitchen AI assistant and should tend to answer in facts and details and that are relevant to the current moment."

Same thing with Alexa. I ask what is the weather outside. She gives me the current weather, what tomorrow looks like, and asks me if I would like an hour by hour breakdown or a full forecast for next week (or something like that).

So then I have to stop her mid-sentence and say, "Alexa, stop." Only for her to misunderstand what I said and say, "Now playing 'On the Rocks' by [artist name]."

I'm in the process of going as analog as I can.

They've spent a lot of money on this technology, and what users want is just not that high a priority.