For this project, I reviewed Kate Crawford's Atlas of AI book, which isn't explicitly a design book. However, it is relevant for designers, especially considering the amount of technology the average person uses on a daily basis. If we are to design for these systems, it is imperative to understand them holistically.
For my process, I read Crawford's book several times so that I may better review it. Much of my understanding of its relevance to design was shaped by the many readings I was introduced to throughout the course.
All that I learned from this project is best represented in the outcome; I'll let it speak for itself.
Confronting the Power Structures of AI
Designers solve problems. We've heard this mantra many times before—and for good reason. As designers, we are often tasked with finding creative and effective solutions for a broad range of either open-ended or nebulous "problems." However, not all solutions turn out to be good ones. When we design, we should thoroughly explore and research not only what could fix our current problem, but also that we are not generating more problems with our interventions.
In an increasingly digital and tech-driven world, we must understand the technologies for which we design. We strive to understand the consequences of our designs: which opportunities they provide and which they deny, who benefits and who may suffer. We research and question not to complicate our process, but so that we can reach solutions that are ethical.
Understanding technologies—particularly machine learning (ML) and artificial intelligence (AI)—can be a challenge. How can we, as designers, begin to understand the complex algorithms and computations of modern technology when the people who wrote the code don't completely understand it? In her book Atlas of AI, Kate Crawford attempts to demystify AI by positing that it is neither artificial nor intelligent.
Crawford is not a designer, and her book does not explicitly address design as a theme. However, Atlas of AI could be incredibly useful in helping those within the design field understand artificial intelligence and the modern technologies with which we work. After all, how can we know we are making ethical design decisions for modern technologies if we do not understand their ramifications?
Using an atlas as a metaphorical framework, Crawford brilliantly allows herself the creative and editorial freedom to explore AI at different scales, locations, and time, making deep connections and allowing us to understand AI more holistically. By using an atlas as a framework, she's acknowledging her own subjectivity and incompleteness in her work. It's also a subtle allusion to one of AI's overarching goals: to map the world.
Atlas of AI is an account of how AI is made, what it is made from, and the forces that give it shape. When you try to visualize artificial intelligence, you may think of ribbons of blue or green code floating on a dark screen or a glowing and intricate web of interlacing nodes. Through her book, Crawford helped me to see AI represented in a way that's more aligned with its reality: rapidly depleting resources, toxic e-waste dumping grounds, scarred landscapes, injured bodies, efficiency over humanity, the degradation of privacy and consent, harmful classifications, the deepening amplification of biases, flattening and misjudging the human emotional experience, and state surveillance from the municipal level to the military.
Crawford argues that AI is neither artificial nor intelligent, but rather that it is both embodied and material. She portrays AI as an industry of extraction and exploitation. She selects her examples masterfully, choosing the most relevant anecdotes and weaving them throughout Atlas of AI so that we can begin to see the underlying systems of power and who these systems serve. All of this she culminates into her last chilling chapter: State.
As I read State, Crawford's description felt surreal. At this point, Crawford has done such a meticulous job visualizing the grossly asymmetrical systems of power. I began to see that Atlas of AI is more than a clever allusion to maps; the Atlases of AI are the people who must maintain and be surveilled by AI. Artificial intelligence is borne on the shoulders of those who suffer from it the most.
From a design perspective, Crawford presents AI as a wicked problem, and she concludes Atlas of AI by approaching it in the same way a designer for social justice would. She argues that to reform AI, we must start with the voices of those AI has most harmed. Rather than allowing AI to continue to be abstracted away, we must focus on the social and environmental harms. We should strive for justice across AI systems.
As designers, we have the responsibility and the privilege of creating more just systems. We endeavor to solve problems—not create them. To do that, we must first understand the powers that shape and drive our digital landscapes. Kate Crawford's Atlas of AI provides the history of technology and AI as industries of extraction and exploitation. We use design every day to shape our world; using an understanding gained from Crawford's book, we can begin to question and push back on AI and technical progress for its own sake.