ending your LLM prompts/queries with “please, thank you” seems to have a 43% improvement over prompts without this (source: trust me bro)
https://www.theregister.com/2025/05/28/google_brin_suggests_threatening_ai/
uh oh am i guilty of misinformation now !!
i’ve kinda wondered if this is true for a while but never actually tried it. i feel mean now.
https://www.theregister.com/2025/05/28/google_brin_suggests_threatening_ai/
uh oh am i guilty of misinformation now !!