AI’s Hidden Material and Human Costs Get a Human Ecology Lens at College of the Atlantic

Beyond the Algorithm: A New Look at What AI Is Actually Made Of

Artificial intelligence is often described as weightless software, a cloud-based intelligence that seems to exist apart from the physical world. A campus talk at College of the Atlantic challenges that assumption by asking a different question: what is AI actually made of? The event, titled “The Stuff AI Is Made Of: A Human Ecology of Artificial Intelligence,” features Dr. Hien Nguyen, the COA Cody van Heerden Chair in Economics & Quantitative Social Sciences, and will examine the material, human, and ecological systems that make AI possible.

Rather than focusing on model performance, benchmark scores, or product features, the presentation is expected to draw attention to the hidden infrastructure behind artificial intelligence: the minerals and metals used in semiconductors, the data centres that consume electricity and water, and the labour force that builds, labels, and maintains AI systems. By positioning AI within a human ecology framework, the talk connects economics, social science, and environmental inquiry in a single critical lens.

The invisible supply chain behind AI

AI systems depend on a vast physical supply chain that is rarely visible to users. The hardware that powers machine learning includes specialized chips made from raw materials such as copper, lithium, cobalt, and rare earth elements. Extracting these materials involves mining, refining, and manufacturing processes with significant environmental and social consequences. Behind those processes are workers in extractive industries, electronics factories, and data centres, many of whom operate under demanding conditions far from the boardrooms where AI strategies are shaped.

The human element also extends to data work. Large language models and computer vision systems require training data that is often cleaned, labelled, and moderated by people. This labour can be repetitive, low-paid, and psychologically taxing, yet it remains foundational to systems that are marketed as automated and intelligent.

• Mining and refining of minerals used in semiconductors and hardware
• Energy and water consumption by data centres that train and run models
• Data labelling, moderation, and maintenance labour that underpins automated systems

Energy, water, and environmental pressure

Training and running AI models is energy-intensive. Data centres that support AI workloads require large amounts of electricity, and cooling those facilities often requires significant volumes of water. Communities that host data centres can face pressure on local power grids and water supplies, while the greenhouse gas emissions associated with energy generation contribute to broader climate concerns. Organizations such as the International Energy Agency have begun tracking the energy implications of artificial intelligence, reflecting growing scrutiny of AI’s environmental footprint.

These costs are rarely itemised when consumers use a chatbot, generate an image, or ask a voice assistant for the weather. The College of the Atlantic event aims to make those connections explicit, treating AI not as an abstract algorithm but as a system with real-world inputs and consequences.

A human ecology lens on technology

College of the Atlantic, located in Bar Harbor, Maine, is known for its interdisciplinary approach to human ecology, which examines the relationships between people and their environments. Dr. Hien Nguyen’s role as the COA Cody van Heerden Chair in Economics & Quantitative Social Sciences places the discussion at the intersection of economic analysis and social science, offering a framework that goes beyond technical explanation to ask who benefits from AI and who bears its costs.

The event page can be found on the College of the Atlantic website, and Dr. Nguyen’s faculty profile provides additional context on the speaker’s background.

The title itself signals a shift: before asking what AI can do, the talk asks what AI is made from — and who and what bears the cost.

A broader debate on responsible AI

The campus talk arrives amid intensifying global debate over AI governance, sustainability, and labour practices. Policymakers, researchers, and advocacy groups have called for greater transparency about the resources required to develop and deploy AI systems. Some companies have published environmental reports or committed to cleaner energy, but critics argue that such measures often capture only a fraction of the full supply chain impact.

By foregrounding the material and human foundations of artificial intelligence, the event offers a counterpoint to narratives that treat AI as infinitely scalable or immaterial. It suggests that a more complete understanding of AI must include the mines, power plants, water systems, and workers that make it function.

For students, faculty, and members of the public, the talk presents an opportunity to consider a familiar technology through an unfamiliar lens. As AI becomes more embedded in daily life, conversations about its hidden costs are likely to become more urgent rather than less.