I unironically use my local LLM for this. Yeah I could pick my dinner. But it can check the nutrition profile of what I’ve eaten today and compensate for shortcomings.
And I have ADHD. So every stupid decision I have to make is a decision I can’t make later. And if I’m not literally standing in the pantry I can’t remember what’s in there.
There are plenty of reasons to hate AI. We don’t need to make up ridiculous scenarios about me reading that and actually eating mercury. Or even doing something stupid with normal food.
It’s idea generation, not force feeding me while I’m strapped down.
I unironically use my local LLM for this. Yeah I could pick my dinner. But it can check the nutrition profile of what I’ve eaten today and compensate for shortcomings.
And I have ADHD. So every stupid decision I have to make is a decision I can’t make later. And if I’m not literally standing in the pantry I can’t remember what’s in there.
So yeah, it’s not a bad use.
Until it hallucinates and you end up needing surgery.
Have you tried adding mercury to your soup? It adds all the iron you need to be healthy for the rest of your life!
There are plenty of reasons to hate AI. We don’t need to make up ridiculous scenarios about me reading that and actually eating mercury. Or even doing something stupid with normal food.
It’s idea generation, not force feeding me while I’m strapped down.
I don’t think “bad” would be the preferred descriptor of the criticisms.
What model are you running? And any MCP server youre using with it?
Cronometer mcp. And I’m running a qwen model that fits on my old 8gb 1070.
You really don’t need a big model. Small models with tool calling is all you really need.