Loadingā¦
25
Unpopular opinion: a guy in a Discord server told me my AI prompts were too polite, and he was right
I used to write prompts to chatbots like I was writing an email to a boss, you know, lots of pleases and could you maybe help me with. Then about 3 weeks ago a guy in a small Discord server for language models quoted one of my prompts and just typed "this is why your answers are mush." I was annoyed at first but he explained that when I hedge and soften everything, the model has no idea what I actually want, so it gives me a generic answer that covers every possible thing. He told me to drop the manners and just state the task plus the format I want, like "summarize this in 5 bullet points for a 12 year old" instead of "I was wondering if you could maybe summarize this for me, thanks so much." I tried it on my grocery list parser and a few work emails and the answers got way shorter and way more useful, like actually usable instead of a wall of fluff. I still say please sometimes out of habit but honestly my whole workflow changed. Has anyone else noticed the softer you ask, the vaguer the reply gets, or is that just how I was prompting?
0 comments
Log in to join the discussion
Log In0 Comments
No comments yet
Be the first to share your thoughts on this discussion.