I only trust LLMS with questions whose answers prove themselves correct or incorrect - so basically code, if it runs and produces the result I was looking for t

I only trust LLMS with questions whose answers prove themselves correct or incorrect - so basically code, if it runs and produces the result I was looking for then great, or where the answer is a stepping off point to my own research on something non-critical like travel. ChatGPT is pretty good at planning travel itineraries, especially if pre promoted with a good description about the groups interests. Beyond that I don’t trust them at all.