The Cost of Being Polite: How “Please” and “Thank You” Waste Energy in Large Language Model Prompts
Qazi Fazli AZEEM
Abstract. Large language models are being embedded in everyday communication, and many users are instinctively applying social norms like politeness to their daily interactions with chatbots. People frequently type phrases such as “please” and “thank you” even though these expressions serve no functional purpose for an artificial system. What appears harmless at the level of a single request becomes more significant at scale because extra words mean extra tokens, extra computation, and therefore extra electricity. Drawing on recent estimates of ChatGPT’s energy use, usage volumes, and user behaviour, this paper develops an exploratory quantitative estimation in which a browser plugin automatically strips polite filler language from prompts. Using a conservative and balanced scenario of a three percent reduction in energy use for polite prompts, the analysis suggests that removing these polite fillers could save approximately 15,075 kilowatt-hours (kWh) of electricity every day. This becomes 5.5 million kWh annually, comparable to the yearly electricity consumption of about 520 average American households. These results do not imply that politeness toward machines must be abandoned entirely. Instead, the paper argues that in a world where the growing number of data centres already account for a significant share of global electricity demand, small linguistic habits deserve attention as part of broader digital sustainability efforts. For a country like Morocco, which is rapidly expanding its digital infrastructure while still heavily reliant on imported fossil fuels, these seemingly minor efficiencies can have a meaningful impact on national energy security and achievable climate goals.
Keywords
Artificial Intelligence, Energy Consumption, Large Language Models, Sustainable AI, Prompt Engineering, Morocco
Published online 4/25/2026, 5 pages
Copyright © 2026 by the author(s)
Published under license by Materials Research Forum LLC., Millersville PA, USA
Citation: Qazi Fazli AZEEM, The Cost of Being Polite: How “Please” and “Thank You” Waste Energy in Large Language Model Prompts, Materials Research Proceedings, Vol. 64, pp 1034-1038, 2026
DOI: https://doi.org/10.21741/9781644904091-128
The article was published as article 128 of the book Energy Futures
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
References
[1] A. Silberling, ChatGPT users send 2.5 billion prompts a day, TechCrunch, 21 July 2025. https://techcrunch.com/2025/07/21/chatgpt-users-send-2-5-billion-prompts-a-day/
[2] A. Sengupta, Sam Altman: Saying “please”, “thank you” to ChatGPT is costing OpenAI “tens of millions of dollars”, Moneycontrol, 20 Apr. 2025. https://www.moneycontrol.com/news/trends/sam-altman-saying-please-thank-you-to-chatgpt-is-costing-openai-tens-of-millions-of-dollars-12998767.html
[3] J. Wilkins, Sam Altman Admits That Saying “Please” and “Thank You” to ChatGPT Costs Tens of Millions of Dollars, Futurism, 19 Apr. 2025. https://futurism.com/altman-please-thanks-chatgpt
[4] International Energy Agency, Energy and AI, IEA, Paris, 2025. https://www.iea.org/reports/energy-and-ai
[5] J. You, How Much Energy Does ChatGPT Use, Epoch AI, 7 Feb. 2025. https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use
[6] United States Energy Information Administration, How much electricity does an American home use?, US EIA, 2024. https://www.eia.gov/tools/faqs/faq.php?id=97&t=3
[7] J. Cockayne, Morocco Renewables Powergen Share Hits Record 25% For 2024, Middle East Economic Survey 68/10, 7 Mar. 2025. https://www.mees.com/2025/3/7/power-water/morocco-renewables-powergen-share-hits-record-25-for-2024/
[8] Reuters, Morocco plans coal power phase-out by 2040, Reuters, 23 Oct. 2025. https://www.reuters.com/sustainability/boards-policy-regulation/morocco-plans-coal-power-phase-out-by-2040-2025-10-23/
[9] Reuters, Morocco to build data centre powered by renewable energy, Reuters, 9 July 2025. https://www.reuters.com/sustainability/boards-policy-regulation/morocco-build-data-centre-powered-by-renewable-energy-2025-07-09/
[10] T. Rehman, D.K. Sanyal, S. Chattopadhyay, How Green are Neural Language Models? Analyzing Energy Consumption in Text Summarization Fine-tuning, arXiv preprint arXiv:2501.15398, 2025. https://doi.org/10.1007/978-981-96-9724-3_47
[11] GreenNLP Project, GreenNLP: Reducing the carbon footprint of natural language processing, University of Helsinki, University of Turku, CSC, n.d. https://greennlp.github.io/
[12] S. Chavali, Prompt Optimization Techniques, Arize AI, 17 Mar. 2025. https://arize.com/blog/prompt-optimization-few-shot-prompting/
