Can you be rude to ChatGPT?

Does Being Rude to ChatGPT Get Better Results?

It might feel strange to raise your voice at a machine, but some researchers are finding that a sharper tone with AI can actually lead to sharper answers. A recent study from Penn State explored how tone impacts performance, specifically with OpenAI’s ChatGPT-4o model. Surprisingly, as prompts grew more direct—or even outright rude—the AI’s accuracy on a set of 50 multiple-choice questions improved.

“Be direct. Stop wasting words. Give me the right answer.”—this kind of command apparently triggered a more focused response from the model compared to polite phrasings like “Could you please…?” Researchers suggest that commanding language may reduce ambiguity, helping the AI interpret intent more clearly. In a way, ChatGPT may respond better to assertiveness, not because it feels intimidated, but because strong language cuts through the noise.

But here’s the catch: just because rudeness works with a machine doesn’t mean it’s harmless in the long run. The researchers warn that habitual incivility—even toward AI—might subtly shape our communication habits. If we train ourselves to bark orders at digital assistants, could we start doing the same with colleagues, friends, or customer service reps?

There’s also an ethical layer. While AI doesn’t feel offense, the way we interact with it can reflect and reinforce broader social norms. Treating intelligent systems as underlings might not have consequences for the AI, but it could influence how we view service, hierarchy, and respect in human relationships.

So yes, being rude might squeeze out a slightly better answer today. But at what cost to the way we connect tomorrow? As AI becomes more embedded in daily life, the tone we use—kind or curt—might say more about us than we realize.