I use Gemini to create images and Meta AI to help me inside Messenger when I'm chatting with students like Jack in Iraq or Quang in Vietnam.
It saves time. It saves money. And as I learned this week, it also saves water.
Wait, AI Uses Water?
Yes. AI doesn't live in the cloud. It lives in huge, hot buildings called data centers.
Those servers get as hot as an oven. To cool them down, companies pump millions of liters of fresh water every day. The water evaporates to remove heat.
According to Li et al. (2025), training GPT-3 can evaporate 700,000 liters of fresh water. The UN University (2025) estimates global data centers could reach 9.3 trillion liters by 2030, a finding also reported by Reuters (2026).
One question uses about 5 drops today. But when we ask 20 questions a day, 30 days a month, those drops become thousands of liters.
In 2024, all data centers used 4.5 trillion liters of water. Experts say it will double by 2030.
Why I Stuck With Just Gemini and Meta AI?
When I learned this, I was glad I already simplified my workflow: 1. Gemini is my workhorse. It's efficient, it connects to Google Docs, and I can batch everything in ONE good prompt: "Make me a 60-minute lesson for a safety engineer about power plant safety discipline, B1 level." One prompt, not five.
2. Meta AI is my quick assistant. Because I'm already in Messenger and WhatsApp with my students, I don't need to open another website. Quick revision, quick paraphrase — done. Less switching, less prompting.
No more opening ChatGPT, Claude, Perplexity, and Canva AI for the same task. Fewer prompts, less server load, less water and energy.
My new rule - If I can't do it with Gemini or Meta AI, I do it myself. And honestly, doing it myself is often better — my students need my experience, not just a perfect worksheet.
The Human Part Costs Zero Water.
AI can generate 10 communication patterns. But only I can say to Jack: "This sentence is good, but in a real safety audit, say it firmer."
That human feedback is free, personal, and sustainable.
I'm not quitting AI. I'm just using it intentionally.
How about you? How many AI tools do you use for teaching?
Let me know below.
Refrerences: Li, P., Yang, J., Islam, M. A., & Ren, S. (2025). Making AI less “thirsty”: Uncovering and addressing the secret water footprint of AI models (arXiv:2304.03271v5). arXiv.
United Nations University Institute for Water, Environment and Health. (2025). Environmental cost of AI's energy use: Carbon, water and land footprints. UNU-INWEH.
Reuters. (2026, June 3). AI to double data centre power and water consumption by 2030, UN researchers say. Reuters.
No comments:
Post a Comment