AI Energy Myth Busted: 'Please' in ChatGPT Prompts Has Negligible Impact
Expert Debunks 'Please' and 'Thank You' AI Energy Myth

Forget what you've heard about typing 'please' to your AI assistant. The widespread belief that using polite language like "please" and "thank you" in ChatGPT prompts significantly drives up energy consumption is largely a myth, according to a leading expert.

The Real Culprit: AI's Insatiable Appetite for Computation

In an original report for The Conversation published on Thursday 15 January 2026, expert Richard Morris explained the science behind AI's energy demands. The core issue is not a few extra words, but the fundamental way artificial intelligence systems operate.

Unlike streaming a film or loading a webpage, which often retrieves pre-existing data, each AI query requires a fresh, complex computation. This makes AI behave more like heavy infrastructure than conventional software. The effect of adding "please" or "thank you" is negligible when compared to the massive energy required for the core task.

Data Centres: The Growing Environmental Footprint

The true environmental impact lies in the data centres that power these AI systems. These facilities already account for a significant and rapidly growing portion of global electricity use. Demand is projected to potentially double by the end of this decade.

However, the footprint extends far beyond just electricity. Data centres also require vast quantities of water for cooling systems and occupy substantial tracts of land. This creates tangible local environmental strains, even when the servers are serving users on the other side of the world.

Beyond Behavioural Tweaks: A Call for Systemic Planning

Experts, including Morris, argue that focusing on minor user behavioural changes is a distraction. While mindful usage is positive, it does little to address the scale of the problem.

The critical need is to integrate AI infrastructure into broader strategic planning for energy, water, and land use. Managing the substantial environmental footprint of AI requires systemic solutions and policy, not just politeness in prompts. This discussion comes amid related news of OpenAI launching GPT-5, its first flagship ChatGPT update in years, which will likely further intensify these infrastructure demands.

In short, you can keep saying 'please' to ChatGPT. The far more pressing conversation is how we power the technology behind it.