[imp0cat]

I get your point, but:

> "If it makes a mistake, it's because I screwed it up. "

Is that really true though with an LLM? I don't think so.

[xtracto]

Oh but it really is. An LLM is no magic wand. It's a model as its name rigthly indicates.

Those of us who have worked in modeling and simulations know two principles of models:

- garbage in, garbage out

- all models are wrong, some are useful.

Point being, usually when "the model is wrong" it's because we put the wrong input (too complicated prompt for the model?) or something similar. Its just a machine, like a meat grinder that we turn then cogs.

[queenkjuul]

I really think LLMs are capable of screwing up good input. They're not deterministic like a meat grinder

[streb-lo]

I think you mean "They're not deterministic, like a meat grinder"