I came across yet another post claiming that AI is sustainable because it consumes less electricity than watching TV.
At first, I smiled. Then I realized I couldn’t remain silent.
This article is for those who don’t settle for slogans. For those who want to face complexity and make informed choices. Because it’s still human intelligence—not artificial—that gives us the tools to do so.
The Fallacy of Convenient Comparisons
Saying that AI is sustainable because it consumes less than television is like saying gas stoves pollute less than coal-burning ones. Technically true, but completely irrelevant.
Sustainability isn’t about comparisons. It’s about systemic coherence: intention, impact, and context.
The Data, in Order. To Truly Understand.
Here are some authoritative sources that help us look beyond the surface:
- European Environmental Agency – 2023 Digitalisation and Sustainability: Unregulated AI accelerates technological obsolescence, increases electronic waste, and intensifies pressure on critical resources. https://www.eea.europa.eu/en/analysis/publications/environmental-statement-2023
- MIT Technology Review – 2023 Training GPT-3 emitted more carbon than 100 cars: Training GPT-3 generated around 552 metric tons of CO₂—equivalent to the annual emissions of over 100 cars. https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/
- Stanford AI Index Report – 2024 AI’s environmental cost is rising despite efficiency gains: Despite improvements, AI’s environmental impact continues to grow: more computations, more energy, more hardware. https://hai.stanford.edu/ai-index/2025-ai-index-report
- MIT Sloan Management Review – February 2024 The Relationship Between Machine Learning and Climate Change: AI can support the climate transition—but only if integrated ethically and intentionally. https://mitsloan.mit.edu/ideas-made-to-matter/climate-change-and-machine-learning-good-bad-and-unknown
- IEA – Electricity 2024 Report Electricity demand from data centres and AI will double by 2026: Within two years, data centers are expected to exceed 1000 TWh per year—more than the total electricity consumption of Japan. https://www.iea.org/reports/electricity-2024/executive-summary
A part of the Story
Yes, it’s also true that:
- AI currently represents less than 1/10,000 of global electricity consumption;
- Domestic water leaks in the US are 30 times more impactful than AI’s water footprint;
- DeepMind has reduced data center energy use by 20%;
- Companies like Line Vision, FIDO Tech, Microsoft, and others are applying AI to boost efficiency and environmental monitoring.
All of this is true. But it’s only half the story.
Because those facts don’t contradict the previous ones—they complete them. And they remind us that technology is not inherently sustainable. It becomes sustainable only when it’s designed, governed, and applied with intention.
Moreover, the positive data shows what is possible—not what is widespread. Solutions exist, but they’re not yet the norm. Success stories are encouraging, but they’re not enough to counterbalance the unchecked growth that, if left unregulated, could amplify inequality and environmental impact.
Without conscious governance, benefits will remain isolated while the costs are shared by all.
Artificial intelligence doesn’t replace our own. It amplifies it.
That’s why we need clarity, not slogans. Vision, not convenience. Critical thinking, not automatic assumptions.
Sustainability is still a human act. A decision. A responsibility.