I still don't trust the non-determinism of current LLMs. I feel like I can't trust the results unless they are very simple ones.

Yeah. Everyone keeps comparing it to higher level languages.

But higher level languages were deterministic and reliable, allowing users to offload the cognitive load of the lower levels to the computer.

With LLMs you can’t fully offload the load, you need to keep a close eye on it which kind of defeats the purpose (IMO)

	▲	jay_kyburz 2 days ago \| parent [-]
		It's not really all that different from trusting some random post on stack overflow. You always needed to be a little skeptical. I'm asking the AI things that are easy to verify, and often ask it to provide web references. It's working well for me. I don't ask it about niche topics. I occasionally ask it about myself or my games and it's always funny. It's a good reminder how wrong the AIs can be.