Do You Have to Be Polite to AI?
Monday, 2026/03/02200 words3 minutes627 reads
The mythology surrounding "prompt engineering" has spawned numerous theories about optimizing AI interactions—from politeness and flattery to threats and role-playing. However, research reveals a more nuanced reality. While a 2024 study found cultural variations in how politeness affected AI accuracy (with Japanese showing decreased performance with excessive courtesy), subsequent research has yielded contradictory results, rendering such techniques largely obsolete.
Jules White, a computer science professor at Vanderbilt University, emphasizes that effective AI communication isn't about "magic words" but rather how you fundamentally structure your requests. Modern LLMs have evolved significantly, becoming more adept at extracting essential information regardless of superficial linguistic variations. Rick Battle, who co-authored the Star Trek study, notes that constant updates to AI systems make previous findings quickly irrelevant.
The most effective strategies focus on practical techniques: requesting multiple options to encourage critical thinking, providing examples to establish context, using iterative questioning to gather information, and maintaining neutrality to avoid confirmation bias. While role-playing can enhance creative tasks, it may encourage hallucinations when seeking factual information. Ultimately, the case for politeness rests not on computational benefits but on philosophical grounds—maintaining courtesy as a habit that shapes one's character, echoing Kantian ethics about how we treat all entities.
