## Introduction: The Misconception of Artificial Intelligence When we talk about artificial intelligence, we often get caught up in the abstract, the ephemeral, and the futuristic. But what if I told you that AI is neither artificial nor intelligent in the way we commonly think? It is profoundly material, rooted in the earth, labor, and the complex web of human relationships. In my book, *Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence*, I aim to map out the true landscape of AI, to show you the physical and social world that underpins this technology. Let's embark on a journey that takes us from the mines of Nevada to the fulfillment centers of Amazon, and from the historical archives of craniology to the modern-day algorithms that shape our lives. ## The Material Roots of AI Imagine driving a van from the heart of Silicon Valley to a tiny mining community in Nevada’s Clayton Valley. This is a journey I took, and it was eye-opening. Here, in the midst of a desolate landscape, lies the source of lithium, a crucial element in the batteries that power our computers and smartphones. The extraction process is anything but clean; it involves destructive environmental practices that starkly contrast with the sleek, high-tech world of Silicon Valley. This dichotomy is not just geographical; it is also economic and social. The wealth disparity between these two places is staggering. While Silicon Valley thrives on innovation and wealth, the mining communities bear the brunt of environmental degradation and health risks. This is the unseen cost of our technological advancements, a cost that is often glossed over in our euphoric embrace of AI. ## The Historical Context: Classification and Inequality To understand the deeper implications of AI, we need to delve into history. Let's visit the Penn Museum in Philadelphia, where rows of human skulls collected by Samuel Morton, a 19th-century American craniologist, are on display. Morton believed he could objectively classify these skulls into five distinct races based on their physical measurements. This work was not just a product of its time; it reflects a broader tendency to categorize and classify, a tendency that persists in modern AI systems. These classifications are far from objective; they impose a social order, naturalize hierarchies, and magnify inequalities. AI systems continue this legacy by classifying the world into fixed categories, often reinforcing existing biases and discriminations. This is not just a matter of flawed algorithms; it is a reflection of the societal structures that these systems are built upon. ## The Global Impact: Labor, Energy, and Climate AI is often celebrated for its efficiency and innovation, but what about the labor and energy that go into creating and sustaining these systems? Let's look at the Amazon fulfillment centers, where human bodies are mechanized in the relentless pursuit of growth and profit. Workers are treated as mere extensions of machines, their lives dictated by the rhythms of automation. The environmental impact is equally alarming. The production and disposal of AI devices involve a massive extraction of minerals and energy, only to have these devices discarded in a matter of years. This is what I call the "deep time" of AI, where the long-term costs of extracting and processing materials are juxtaposed with the short-term benefits of technological convenience. ## Enchanted Determinism: The Paradox of AI In the conclusion of my book, I introduce the concept of "enchanted determinism." This paradox highlights the contradictory nature of AI systems: they are both "enchanted" in their ability to perform tasks beyond our explicit instructions and deterministic in their adherence to programmed rules. However, this duality is problematic when we expect these systems to make high-stakes decisions that require discretion. An algorithm cannot both follow deterministic instructions and incorporate randomness; it cannot be both safe and ethical in the way we expect human decision-makers to be. This paradox underscores the need for a more nuanced understanding of AI, one that acknowledges its material and social roots rather than treating it as a magical solution to our problems. ## Mapping the Future: A New Perspective on AI The title *Atlas of AI* is not just a metaphor; it is a call to action. An atlas allows us to zoom in and out, to see the big picture and the minute details. It encourages us to think about AI at different scales: from the global dominance of a handful of tech companies to the everyday impacts on our lives. By mapping AI in this way, we can begin to see the interconnectedness of our world. We can understand how the extraction of minerals in one part of the world affects the labor practices in another, and how these practices, in turn, shape our relationships with technology and with each other. This is not just about critiquing AI; it is about transforming our understanding of it. It is about recognizing that AI is not a neutral or objective technology but a deeply material and social construct that reflects and shapes our world. As we move forward, it is crucial that we attend to the commitments we are making as we develop and deploy AI systems, commitments that will have far-reaching consequences for our planet and our humanity. In the end, *Atlas of AI* is not just a book; it is a journey through the physical and social landscapes that underpin our technological advancements. It is a reminder that AI is neither artificial nor intelligent in the way we think, but it is profoundly real in its impacts on our world.

Here are the key insights from Kate Crawford's book, *Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence*: ## The Material Roots of AI - AI is profoundly material, rooted in the extraction of resources such as lithium, which involves destructive environmental practices and significant health risks for mining communities. ## Economic and Social Disparities - There is a stark wealth disparity between the high-tech world of Silicon Valley and the mining communities that supply the necessary resources, highlighting the unseen costs of technological advancements. ## Historical Context: Classification and Inequality - Historical practices of classification, such as those by Samuel Morton, reflect a broader tendency to categorize and classify, which persists in modern AI systems, reinforcing existing biases and inequalities. ## Global Impact: Labor, Energy, and Climate - AI systems rely on labor that is often mechanized and exploitative, as seen in Amazon fulfillment centers, and have a significant environmental impact due to the extraction and disposal of minerals and energy. ## Enchanted Determinism - AI systems exhibit a paradoxical nature, being both "enchanted" in their ability to perform beyond explicit instructions and deterministic in adhering to programmed rules, which is problematic for high-stakes decision-making. ## Mapping the Future - The book advocates for a nuanced understanding of AI by mapping its impacts at different scales, from global tech dominance to everyday life, highlighting the interconnectedness of the world and the need for ethical and sustainable AI development. ## Critique of Big Tech - The book criticizes Big Tech for exploiting labor and resources, and for the amplification of inequality and surveillance through AI systems. ## Data and Classification Issues - AI systems often use datasets built without consent and trained on biased or incorrect data, which can lead to inaccurate and discriminatory outcomes, such as in the case of ImageNet and Amazon's job applicant classification project. ## State Surveillance and Decision-Making - The book discusses the use of AI in state surveillance, intelligence gathering, and decision-making, highlighting issues such as the involvement of private corporations and the ethical concerns surrounding these practices. ## Call to Action - The book is a call to action, urging a reevaluation of AI technologies to ensure they are ethical, sustainable, and equitable, and emphasizing the need to attend to the commitments made in developing and deploying AI systems.