A seemingly casual remark online recently turned into an unexpected conversation around the financial impact of politeness on artificial intelligence models. Responding to a user’s question on the social platform formerly known as Twitter (now X), OpenAI CEO Sam Altman noted—with a playful yet thought-provoking tone—that people including polite expressions such as “please” and “thank you” in interactions with AI chatbots have cost his company “tens of millions of dollars,” though he humorously described it as money well spent.
Despite Altman’s nonchalant framing, his comment prompted wider discussion about the practical implications of politeness when communicating with generative AI models like ChatGPT. While engaging in courteous exchanges with a program might initially seem harmless, some observers questioned whether the added words serve as an unnecessary drain on computing resources and electricity.
However, Kurt Beavers, director of Microsoft Copilot’s design team, countered this viewpoint, stating that gentle and polite language significantly influences how AI tools respond. According to Beavers, when AI models detect politeness, they tend to reciprocate similarly courteous language. This suggests that courteous interactions aren’t merely about personifying software or an irrational fear of advanced AI systems—they can genuinely enhance the quality of the engagement.
While politeness has clear benefits, others also noted that occasionally using strong language or profanity can serve practical purposes in eliciting certain responses or bypassing AI-generated content restrictions.
