.svg)
Does being polite to AI improve your results or waste energy? We weigh the environmental cost against the performance benefits of "politeness prompting."
TL;DR
In the early 2000s, the world was gripped by a different kind of linguistic panic: ‘text speak’. Critics warned that a generation using u instead of you would lead to the total degradation of the English language. Fast forward to 2026, and while text speak didn't destroy society, we are facing a new social dilemma: do we need to be polite to our machines?
According to our latest SalesAPE survey, 75.7% of people find it positive or neutral when an AI identifies itself as a bot immediately. We clearly know it’s a machine, yet many of us still find ourselves typing please and thank you. Is this just basic human instinct, or is it a waste of processing power?
Did you know…
Between intense energy consumption, massive water usage for cooling, and rising electronic waste. Data centers, needed for AI model training and inference, could consume 4% of global electricity.
One of the most common arguments against AI is its carbon footprint. So from an environmental standpoint, every word we type carries an ecological cost. AI systems process text incrementally, meaning longer prompts require more computation. A study found that models using internal reasoning can produce up to 50 times more emissions when asked to provide long, detailed responses versus concise ones.
While a single ‘please’ won't melt an iceberg, it does add a tiny metabolic load to a system that is already doubling its energy demand every few years. If every one of the billions of daily AI queries includes unnecessary niceties, the cumulative energy cost is no longer marginal. However, there is a flip side: Politeness Prompting actually works.
Because AI is trained on human data, it associates polite, well-structured queries with higher-quality information. In essence, if you bark orders, you might get a half-baked response. If you treat it like a helpful colleague, you’re more likely to trigger the AI’s digital empathy, resulting in clearer and more accurate output.
Think of it like this, if someone came up to you on the street and asked for directions to the nearest ATM, your response would reflect their request. If they demanded ‘where’s the nearest ATM”’ at you in a sharp tone, you might just tell them it’s down the street and vaguely point. If someone said ‘excuse me, would you mind telling me where I can find the nearest ATM?’ you’d be more inclined to offer more detailed directions and be polite in return.
For many autistic users, the rise of AI is a social relief. Traditional human interaction is laden with complex social cues and unwritten manners that can be exhausting to navigate.
AI provides a safe, non-judgmental space where the social tax is zero. Research shows that autistic individuals often respond positively to AI and robots precisely because their behaviors are predictable and lack complex social pressure.
Whether it's practicing social scenarios without fear of rejection or simply being able to ask a direct question without the politeness padding, AI is becoming a powerful tool for neurodiverse accessibility. It isn't about degrading social skills; it's about providing an alternative way to communicate that doesn't require masking.
Then, of course, there’s the old cliché: ‘I’m nice to Siri now so she remembers me when the robots take over.’ While it's a popular joke, it points to a deeper truth about human psychology. Our manners are a reflection of us, not the machine.
If we stop being polite to the thing we interact with for eight hours a day, does that bleed into how we treat our real-life colleagues? A study by ResearchGate found that politeness toward AI tends to decline over time, eroding faster than in human-to-human interactions.
Just as the text-speak era left us with a generation that sometimes struggles with formal grammar, the blunt AI era might leave us with a generation that forgets the social graces that make human collaboration work.
At the end of the day, niceties aren't required for the machine, but they are often beneficial for the results.
Yes, but not because the AI has feelings. LLMs are trained on human conversations where polite language is usually associated with better context and more detailed information. Politeness Prompting can reduce hallucinations and bias by mirroring the high-quality data the AI was trained on.
While a single please is negligible, the length of the response is the real energy driver. Polite prompts that encourage the AI to be more conversational or detailed can increase the energy usage per query. To be green, it is better to ask for concise bullet points rather than a polite, detailed essay.
It’s a double-edged sword. While AI provides a safe space for neurodiverse children to practice social cues, some experts worry that for neurotypical children, a machine that fulfills every demand without a please could lead to a decline in real-world empathy. The consensus is that AI should supplement, not replace, human social interaction.
It’s called Anthropomorphism, the human tendency to attribute human traits to non-human things. Because AI uses "I" and communicates in a natural voice, our brains often bypass the fact that it's code and treat it as a social entity. It's a testament to how human-like these models have become.