30-Second Book Review No.1: Kate Crawford, Atlas of AI
Monday, March 4, 2024 at 7:54AM 30-Second Book Review
No.1 in an occasional series:
Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (New Haven and London: Yale University Press, 2021).
Crawford’s 'seminal' book Atlas of AI is a classic example of a work that emphasizes how media is ‘made from natural resources, fuel, human labor, infrastructures, logistics', while leaving its own media-materiality masked.
It may be the case that, as Crawford pointed out in an article in Nature last week:
Generative AI systems need enormous amounts of fresh water to cool their processors and generate electricity. In West Des Moines, Iowa, a giant data-centre cluster serves OpenAI’s most advanced model, GPT-4. A lawsuit by local residents revealed that in July 2022, the month before OpenAI finished training the model, the cluster used about 6% of the district’s water. As Google and Microsoft prepared their Bard and Bing large language models, both had major spikes in water use — increases of 20% and 34%, respectively, in one year, according to the companies’ environmental reports.
But, as The World Counts shows, it's also the case that:
In the USA, Japan, and Europe an average person uses between 200 and 250 kilos of paper every year. In India the figure is 5 kilos, and in some countries it is less than 1 kilo.
Producing 1 kilo of paper requires 2-3 times its weight in trees. If everyone used 200 kilos of paper per year there would be no trees left.
It takes between 2 and 13 liters of water to produce a single A4-sheet of paper, depending on the mill. The pulp and paper industry is the single largest industrial consumer of water in Western countries.
(Talk about 'denial' and 'the elephant in the room'.)
Of course, there are any number of legacy authors producing new theories of the politics of technology in old ways.
If we want to be more hospitable, then, we can follow Crawford’s lead when she writes in her Atlas of AI about the importance of understanding ‘how AI is fundamentally political’. Taking her ideas about a ‘multitude of interlaced systems of power’ to their logical conclusion, we can see Crawford as encouraging us to ask: ‘what is being optimized, and for whom, and who gets to decide’ with respect to the ‘abstracting away’ of the hegemonic, authoritative, ‘material conditions’ of the making of her own 336 page paperback book as well?