Being mean to ChatGPT can boost its accuracy, but scientists warn you may regret it
Bossing around an AI underling may yield better results than being polite, but that doesn’t mean a ruder tone won’t have consequences in the long run, researchers say.

TL;DR
- A Penn State study found ruder prompts led to higher accuracy from ChatGPT's 4o model on multiple-choice questions.
- Ruder responses yielded 84.8% accuracy, four percentage points higher than very polite responses.
- Using insulting or demeaning language may have negative effects on user experience, accessibility, and inclusivity.
- Later research by the same scientists found prompt tone can affect AI computing costs by as much as 44.3%.
- Rude, threatening language elicits shorter responses, while polite inquiries produce longer answers incurring greater computational cost.
- Anthropic has moved to prohibit sustained cruelty toward its AI models.
- The recommended interaction style for AI is neutral leaning towards firm to optimize computation cost and result quality.