The AI Etiquette Debate: Why Politeness Matters in Prompt Engineering

The question of whether to say 'please' to an artificial intelligence is moving from a lighthearted social media trend to a serious topic of study within the field of Human-Computer Interaction (HCI). As large language models (LLMs) become more integrated into daily workflows, the way humans communicate with these systems is evolving. While it is technically true that an AI does not have feelings or a sense of social obligation, the data on which these models are trained—vast swaths of human-generated text—is deeply rooted in social norms. Consequently, the 'tone' of a prompt can have measurable effects on the accuracy, tone, and thoroughness of the response.

From a technical perspective, the efficacy of politeness in prompting is often linked to the way models are trained using Reinforcement Learning from Human Feedback (RLHF). During this process, human trainers rank responses based on quality, clarity, and helpfulness. Because human trainers often associate polite, well-structured requests with high-quality interactions, the models may inadvertently learn that 'polite' input contexts should be met with more professional or detailed 'polite' output contexts. Some researchers have even identified a phenomenon known as 'emotional prompting,' where adding phrases like 'this is very important for my career' or 'please do your best' can nudge the model toward higher performance benchmarks, effectively acting as a weight on the model's attention mechanism.

“

As large language models (LLMs) become more integrated into daily workflows, the way humans communicate with these systems is evolving.

However, the debate isn't purely technical; it is also psychological. Many experts argue that maintaining social graces with AI is less about the machine's 'feelings' and more about the user's habits. There is a growing concern among sociologists that if users become accustomed to barking short, rude commands at digital assistants, those behavioral patterns could bleed into real-world human interactions. By maintaining a standard of politeness with AI, users may be reinforcing their own positive social habits. This 'habituation' argument suggests that treating AI with respect is a form of psychological self-maintenance in an increasingly automated world.