To say that artificial intelligence (AI) is all the rage today in many circles of finance, business strategy and politics is likely not an overstatement. The share values of publicly-traded firms in the AI space continue to surge. Corporate boardrooms are having increasingly deeper discussions about how to make AI a more integral part of their organizations. Elected officials and bureaucrats are examining how well current regulatory regimes work – or do not – when it comes to AI. Moreover, futurists and prognosticators fill columns, podcasts and blogs daily with tales of AI taking over whole industries; and, possibly one day, humanity itself.
In this maelstrom of fact, fiction and hyperbole stands Kate Crawford and her seminal work Atlas of AI. The book is an in-depth analysis of the tangible political and cultural forces behind the amalgam of technologies that have come to be known as “Artificial Intelligence”. And, while Crawford readily acknowledges the power of AI to do many things, she reminds the reader of the vast amount of natural and human resource contributions that allow AI to even function in the first place.
Crawford begins by exploring the creation of intelligence and the inevitable traps that exist in the process. Fundamental to AI, she asserts, is the notion of classification. Classifying something reduces its totality; and, marginalizes it into a new, predetermined and limiting reality. Of course, those who set the classifications and their criteria are the ones who possess the power to make or break. Herein lies a central theme of the book: the chief architects of AI and those who seek to control it today are a small cadre of unaccountable military and economic elites who desire ever-greater surveillance of human society.
Crawford deftly explores the role of America’s military-industrial complex in building and maintaining the current AI infrastructure, while reminding the reader that commercial applications of AI are still largely dependent on the permission of the original gatekeepers. As she observes, the US Government’s AI strategy seeks “national control and international dominance to secure military and corporate advantage.” Recently, the AI intersection of the military and commercial worlds can be easily observed by the tens of billions of dollars of government contracts given to Amazon, Microsoft and the like for cloud storage services.
Beyond the disturbing concerns behind AI’s creation and continued existence, Crawford details the sheer volume of inputs needed to make it work: an unrelenting need for ever-more rare earth minerals (with all of the environmental and human damage that accompanies their mining); massive amounts of water needed to cool server farms; huge quantities of coal and natural gas required to provide the electricity to keep everything running; and cheap human labor to clean-up and fix algorithms one click at a time. All of this adds up to reveal how AI is simply an insatiable monster that consumes more and more scarce resources.
While not completely against AI in all its forms, Crawford’s book is a warning to those who think a panacea awaits as the commercialization of AI unfolds. “Addressing the foundational problems of AI and planetary computation requires connecting issues of power and justice: from epistemology to labor rights, to resource extraction to data protections, racial inequity to climate change.” She could not be more right.
