There is a growing debate online suggesting that if you want to save the planet, you should stop being polite to your AI. The logic is simple: every extra word—like “please” or “thank you”—requires more computational power, which in turn consumes more electricity.
While technically true that longer prompts require more processing, this focus on “prompt etiquette” is a distraction from the much larger, more systemic environmental challenges posed by artificial intelligence.
The Myth of the “Polite” Prompt
It is true that AI models process text incrementally. Every additional word necessitates a small amount of extra computation. OpenAI CEO Sam Altman has noted that at the scale of billions of prompts, these micro-costs add up to significant operating expenses.
However, the idea that individual politeness affects the planet is a stretch. The energy used to process a few extra words is negligible compared to the massive amount of power required to run the data centers themselves. The real issue isn’t how we phrase our questions, but how much we use these systems.
Why AI is Different from Traditional Software
To understand the environmental impact, we must distinguish AI from the digital services we have used for decades.
- Traditional Services: When you stream a video or open a document, the energy cost is largely “front-loaded.” The data already exists; the system is simply retrieving and delivering it.
- Artificial Intelligence: AI operates through inference. Every time you ask a question, the model performs a fresh, intensive computational pass to generate a unique response.
Because every query triggers new computation, AI behaves less like a digital library and more like heavy infrastructure. Usage translates directly and immediately into energy demand.
The Growing Footprint: Energy, Water, and Land
The scale of this demand is moving from marginal to massive. The environmental footprint of AI is not just about electricity; it is a multi-resource challenge:
- Electricity: The International Energy Agency warns that data center electricity demand could double by the end of this decade.
- Water: Data centers require vast amounts of water for cooling their high-density computing hardware.
- Land and Materials: Building and maintaining the physical infrastructure requires significant land use and raw materials.
These impacts are often felt locally. For example, in countries like New Zealand, which relies heavily on renewable hydro-power, large data centers can strain the local grid. During dry years when water levels are low, the electricity used to run servers is electricity that cannot be used for other essential social needs.
A Shift in Perspective: From Software to Infrastructure
The current debate often treats AI as an “immaterial” digital service—something that exists in a cloud, detached from the physical world. This is a mistake. AI is a physical presence that imposes a “metabolic load” on our existing systems.
When we view AI through a “systems lens,” we see that energy, water, and land are tightly coupled. A spike in AI demand doesn’t just affect the power grid; it affects water availability and land-use planning.
“Focusing on small behavioral tweaks, such as how prompts are phrased, distracts from the real structural issues.”
Conclusion
The obsession with whether we should be polite to ChatGPT is a signal that people intuitively sense AI has a physical footprint, even if they lack the technical language to describe it. To manage this technology sustainably, we must move past “prompt etiquette” and start integrating AI infrastructure into our broader global planning for energy, water, and land use.
