n
2

Prompt engineering saved me 3 hours, but is it a crutch?

Spent 4 hours last week fixing a Python script that kept returning garbage from an LLM API. Turns out the prompt was too vague, no examples, no format spec. Rewrote it with 3 few-shot examples and a JSON schema, fixed in 45 minutes. So I get the hype around prompt engineering, it's real. But now I worry, are we just masking bad model design with clever wording? My team in Seattle argues we should push for better fine-tuning instead. Which side's right, or do we need both?
0 comments

Log in to join the discussion

Log In
0 Comments

No comments yet

Be the first to share your thoughts on this discussion.