The Industrial Revolution propelled humanity into a new age, of prosperity for some, and pollution for all.
Coal, oil, and steel built empires, but also scarred the planet. Only after securing their wealth did leading nations start setting environmental rules, expecting latecomers to comply with standards they themselves ignored during their ascent.
Today, we seem to be witnessing a new form of déjà vu; this time driven by Artificial Intelligence.
The Invisible Cost of AI’s Rise
Behind the dazzling promises of AI – smarter machines, new efficiencies, groundbreaking discoveries- hides an uncomfortable truth:
AI consumes enormous amounts of energy, water, and rare minerals, accelerating environmental degradation in ways that are still widely underappreciated.
Here are a few facts that illustrate the hidden cost:
Energy consumption: Training a single large AI model (like GPT-3) can emit as much carbon as five average cars during their entire lifetimes. (Source: University of Massachusetts Amherst, 2019)
Water usage: Training and operating AI models requires immense cooling systems for data centers. According to research from UC Riverside, training a single GPT-3 model could consume over 700,000 liters of fresh water; enough to fill an Olympic-sized swimming pool.
Rare minerals and mining: AI hardware — GPUs, TPUs, chips — depends heavily on materials like lithium, cobalt, and rare earth elements. Mining these materials leads to ecosystem destruction, toxic waste, and human rights issues, notably in the Democratic Republic of Congo for cobalt extraction.
Nuclear and fossil fuel dependencies: Because renewable energy capacity is still insufficient, many new AI data centers are powered by fossil fuels or by newly incentivized nuclear plants, raising questions about what kind of energy future AI is truly supporting.
Winners Write the Rules (Again)
As rich countries double down on AI supremacy, they do so largely without environmental accounting.
Their infrastructures are growing at breakneck speed; Microsoft, Google, Amazon, and Meta are planning hundreds of new data centers globally.
Once AI becomes the indispensable infrastructure of tomorrow’s economy, these same countries will likely introduce “Green AI” standards, “ethical mining” regulations, and strict compliance frameworks.
Just as they once imposed environmental rules after decades of unchecked industrialization.
Emerging economies, meanwhile, will face the same double bind:
Struggling to catch up technologically, While being judged or sanctioned for practices that the pioneers themselves ignored when they needed speed over sustainability.
What Could We Do Differently? Learning from Small Is Beautiful
In 1973, economist E.F. Schumacher wrote Small is Beautiful, warning against the blind pursuit of bigger, faster, and more; at any cost.
He advocated for “economics as if people mattered” sustainable development, local resilience, and appropriate technology tailored to human and environmental needs.
His warnings feel uncannily relevant today.
Instead of another century of industrial gigantism, we could reimagine AI through the lens of Small Is Beautiful:
1. Build AI Sustainability Standards before global domination
International institutions like the UN, OECD, or ISO should set sustainability benchmarks now, not after the damage is done.
2. Tax the Externalities Upfront
Introduce taxes or levies based on the carbon emissions and water consumption of AI operations; true costs must be visible from the start.
3. Encourage Localized, Smaller-Scale AI
Promote the development of specialized, efficient AI models that solve real problems locally, rather than chasing planetary domination through ever-larger language models.
4. Public Investment in Green AI Research
Governments should invest in low-energy AI architectures, valuing appropriateness over mere scale.
5. Ethical Resource Supply Chains
Ensure rare mineral sourcing respects human rights and environmental balance, not just corporate profit margins.
“Any intelligent fool can make things bigger and more complex.
It takes a touch of genius — and a lot of courage — to move in the opposite direction.”
— E.F. Schumacher
History Doesn’t Repeat, But It Rhymes
AI holds immense promise — from healthcare breakthroughs to climate modeling.
But the way we are racing into this new era looks unsettlingly like the Industrial Revolution redux: exploit first, regulate later; grow first, reflect later; dominate first, moralize later.
If we don’t acknowledge this pattern now, we risk repeating the same mistakes — this time at an even larger, more irreversible scale.
As we build the future with AI, the real question isn’t just how powerful it becomes,
It’s whether we dare to choose a different path.
Sources:
Strubell et al., Energy and Policy Considerations for Deep Learning in NLP (2019)
Patterson et al., Carbon Emissions and Large Neural Network Training (Google Research, 2021)
Li et al., Making AI Less “Thirsty”, UC Riverside (2023)
The World Bank, Mineral Production and the Green Energy Transition (2020)
E.F. Schumacher, Small is Beautiful: A Study of Economics As If People Mattered (1973)
Learning for Performance 