I’m a strong advocate of AI use in education. I’ve spent years exploring how AI can support teaching, enhance learning experiences, and simplify daily tasks for educators. But this enthusiasm should not blind us to the broader consequences of these technologies especially their ecological impact.
I do acknowledge that AI has the power to transform education for the better. At the same time, it’s essential to remain critically aware of the material costs that power these tools. Recently, I revisited Atlas of AI by Kate Crawford, a book that offers a sobering look at the environmental and social implications behind the sleek interfaces of artificial intelligence.
As I revisited my highlights from the book, I was struck by the sheer scale of the hidden infrastructure: from massive energy consumption and data center pollution to mineral extraction and e-waste. I thought it might be helpful to turn these insights into a visual format that educators can use in their own classrooms.
So I created a poster that outlines some of the key ecological impacts of AI. Feel free to download it and share it with your students to spark meaningful conversations about the ethics and sustainability of technology in our lives.
The Hidden Ecological Cost of AI
While AI is often portrayed as sleek, seamless, and immaterial, the reality beneath the surface tells a different story. Behind every chatbot, recommendation engine, or intelligent assistant lies an enormous physical footprint, one that spans continents and ecosystems.
From the energy it takes to train large models to the minerals mined for chips and servers, the ecological cost of AI is both vast and often hidden. Below, I break down some of the key environmental impacts associated with the development and deployment of AI technologies, drawing inspiration from Kate Crawford’s Atlas of AI.:
Massive Energy Consumption
Training today’s large-scale AI models, particularly those built on deep learning architectures, requires immense computational power which translates into equally immense energy demands. These energy needs are typically met by electricity grids still heavily reliant on fossil fuels, resulting in significant carbon emissions.
Carbon Footprint of AI Models
The carbon impact of training advanced natural language processing models is staggering. One well-cited study found that a single large NLP model could emit over 660,000 pounds of CO₂ during its training phase alone. That’s equivalent to the lifetime emissions of five average gas-powered cars. And this number only reflects training not the repeated costs of deployment, fine-tuning, or inference across billions of devices.
Mineral Extraction
The development of AI is heavily dependent on rare earth elements and metals like lithium and cobalt, which are essential for producing semiconductors, batteries, and computing hardware. These resources are often extracted through mining practices that wreak havoc on local ecosystems and communities. Regions like Nevada (USA), the Democratic Republic of the Congo, and Bolivia are some of the key sites where this environmental degradation unfolds, raising deep ethical and sustainability concerns about the true cost of AI’s growth.
E-waste Pollution
As AI technologies evolve, so does the hardware supporting them leading to a rapid turnover of devices, servers, and storage units. When these components are discarded, they become part of a growing stream of electronic waste. Much of this toxic e-waste ends up in countries with weak environmental protections, such as Ghana and Pakistan, where it poses serious health risks to local communities and contributes to long-term soil and water contamination.
Data Centers and Pollution
Behind the smooth operation of AI tools are sprawling data centers that consume vast amounts of electricity to power servers and maintain cooling systems. These facilities often draw from non-renewable energy sources, making them major contributors to air and water pollution. Their environmental impact goes largely unseen by end-users but plays a significant role in AI’s overall carbon footprint.
Shipping and Supply Chains
The creation and maintenance of AI systems depend on a complex global supply chain that spans continents. This includes the extraction of raw materials, manufacturing of chips, and distribution of hardware through container shipping. These ships burn heavy fuel oil, producing more CO₂ emissions annually than entire countries. The environmental toll of simply moving AI infrastructure around the globe is immense and often overlooked.
Depletion of Nonrenewable Resources
AI systems are built on the backs of nonrenewable materials, metals, minerals, and fossil-derived plastics that took billions of years to form. Each new device or server contributes to the depletion of these finite resources. As the demand for AI grows, so does the pressure on already-strained ecosystems, accelerating an unsustainable cycle of extraction and exhaustion.
Hidden Infrastructure
Though AI is often sold as “cloud-based” or “virtual,” the reality is deeply physical. It’s powered by massive server farms, underwater cables, mining operations, and shipping networks. This hidden infrastructure is rarely part of public conversations about AI, yet it constitutes the backbone of every algorithm and application. Recognizing the material reality behind digital intelligence is crucial for understanding its true environmental cost.
Here is the DOWNLOAD link for this poster!
Conclusion
The remarkable rise of AI has brought with it transformative potential across education, research, healthcare, and nearly every facet of modern life. Yet, behind the promise of innovation lies an ecological footprint that is often overlooked. From the immense energy demands of training large models to the mining of rare earth minerals and the toxic legacy of e-waste, AI is anything but immaterial. As educators, researchers, and technologists, we must confront these hidden costs and incorporate them into how we evaluate, use, and teach about AI. Embracing AI doesn’t mean ignoring its impacts, it means engaging with them responsibly, advocating for sustainable practices, and ensuring that our technological progress does not come at the expense of the planet.
