alt Hacker News> It is impossible to duplicate results with an LLM.
Humans are called experts in many domains whereby they cannot achieve duplicate results. In fact, in what domain can an expert do that?
Nobody crosses the same river twice.
> It is impossible to duplicate results with an LLM.
Humans are called experts in many domains whereby they cannot achieve duplicate results. In fact, in what domain can an expert do that?
Nobody crosses the same river twice.
If you ask me to write an email, all 3 variations will be more or less the same, except it will almost always improve with each iteration. 3 cracks at a prompt is 3 separate, unrelated attempts. None of them informs the other. In fact, you run the risk of making it worse if you include previous versions.