Thirsty AI, Viral Distractions, Burning Planet
From Joe Biden to Taylor Swift: What Are We Paying Attention To?
Isn’t it interesting when you finish a book and then discover that the author is a grandchild of JFK, a former U.S. president? I’ll get back to that later. For now, let me share a concern that has been bothering me for a while.
During a climate crisis and a world on fire by warming, warfare and wildfire, mainstream and social media often distract us with the trivial matters. Breaking news in politics, showbiz, and sports constantly shocks us and diverts our attention from the issues that our lives depend on. Consequently, many media outlets — especially podcasters and YouTubers — cover these topics hour after hour while neglecting coverage of pressing environmental issues.
The latest attention-grabbing news, at the time of writing, is the health condition of Joe Biden, the former president of the U.S., who has been diagnosed with an aggressive form of prostate cancer, characterized by a Gleason score of 9 (Grade Group 5). It might not be surprising that an 82-year-old man faces serious health issues (of course, it’s sad to hear of anyone’s suffering), but this disclosure came as a bombshell for many people who are stunned, asking how a seriously ill president was motivated to run for reelection. Even more mysterious is why the U.S. administration kept Americans and the world in the dark, especially while he was in office–what happened to transparency?
In the content economy era, when social media feeds and news outlets are frequently dominated by political scandals and celebrity crises, no surprise that tens of billions of dollars are poured into a ‘green lie’ or ‘greenwashing’, yet go unnoticed.
An investigation, according to , reveals that European ‘green’ funds — such as Sustainable Global Stars and Europe Climate Pathway — hold over $33 billion in oil and gas investments, despite their so-called eco-friendly branding and the role of fossil fuels in the climate crisis.
Similar to Taylor Swift’s dating life, I doubt this news has gone viral. Many real stories concerning the health of planet Earth are buried beneath the buzz. The AI’s inconspicuous footprint is yet another example.
Thirsty AI and Our Negligence
A lesser-discussed but growing issue in the public domain is the environmental impact of generative AI, such as ChatGPT, DALL.E, Manus and others.
Training AI models and processing user requests continuously require substantial amounts of energy and water for data servers. Researchers from the University of Colorado and the University of Texas have tried to quantify how thirsty AI is. In a study titled , they estimated that a single 100-word response from ChatGPT-4 consumes around 519 millilitres of water and 0.14 kilowatt-hours (kWh) of electricity when run on an average U.S.-based data centre. Extrapolating from this, the figures are staggering- and YES! The actual numbers would vary depending on factors, such as the location of a data centre, the energy source (fossil fuels or renewable) and user prompts.
I asked ChatGPT about its water consumption. It referred to the article above, adding: ‘The estimate that training GPT-3 consumed over 700,000 litres (approx. 184,000 gallons) of water refers to a single training session, not an annual figure. This water was primarily used for cooling the data centers’ servers during the training process. To put this into perspective, that amount of water is roughly equivalent to the annual water consumption of five average U.S. households.’
Experts writing in Making AI Less Thirsty highlighted that by 2027, global AI demand could lead to 4.2–6.6 billion cubic meters of water withdrawal annually. This volume is substantial, exceeding half of the U.K.’s total annual water withdrawal.
Again, I asked ChatGPT about its carbon footprint. Here is its answer:
✅ Per interaction (prompt + response):
- Running a single ChatGPT prompt can consume 2–4 watt-hours (Wh) of electricity, depending on complexity and model size.
- That translates to about 1–5 grams of CO₂ per prompt (depending on energy source: coal vs. renewables).
✅ At scale (millions of users per day):
- ChatGPT gets ~100 million users monthly, likely generating hundreds of millions of prompts daily.
- Assuming 100 million prompts/day × 3g CO₂ each =
≈ 300 metric tons of CO₂ per day.
🧠 Context:
- 300 metric tons of CO₂/day ≈ emissions from burning ~33,000 gallons of gasoline.
- Still small compared to entire sectors (like transportation or manufacturing), but significant as AI use scales.
reports that ChatGPT attracts approx. 400 million weekly users, equivalent to roughly 57 million per day. If each submits five prompts a day, that means nearly 286 million responses. By that logic, ChatGPT’s daily footprint would amount to around 148 million litres (39 million gallons) of water and 40 million kWh of electricity. If we add all generative AI to the calculation, the results would be eye-watering!
Despite these alarming figures, users remain unconcerned. According to a user perception survey, most participants, while acknowledging ChatGPT’s size and likely high energy consumption, do not actively consider its environmental impact during use and are unaware of its water consumption, highlighted in a 2024 bachelor’s , submitted to a Swedish university and published on the DiVA academic portal.
Alarms for Sleepers?
Prior to the public release of generative AI, concerns were raised about the potential problems arising from internet usage. I first encountered this information in the opening chapter of a book exploring the hidden environmental impact of the internet. Subsequently, I realized that the author was an environmental journalist at The New York Times. Initially, I was intrigued by her research and sharp writing. Then, I discovered something even more surprising — she’s the granddaughter of JFK and Jacqueline Kennedy.
In a time when legacy often influences public performance, Tatiana Schlossberg has perhaps taken a quieter route, focusing on the environmental costs of our consumption — something that we rarely question. She cautioned through her book Inconspicuous Consumption– subtitle: ‘The Environmental Impact You Don’t Know You Have that ‘as the internet grows in size–measured in bytes– and by the number of people who use it and the network we build, we have to make sure that it grows sustainably and responsibly. Data centres are a good example of why we need to do that, and some of the solutions in that area might provide a good road map to how it might get done.’
Now, here we are, buried under the digital noise and AI, in many shapes and forms, thriving and promising a brighter future while devouring our –and other species’– fresh water resources and emitting greenhouse gases, not to mention the environmental impacts of cryptocurrency mining. The question isn’t whether these technologies will shape our future — they already are. The question is: how can we lessen their negative impacts and allow the Earth to recover?!
