Comment Editor Deborah Solomon unravels the pervasive and unthinking use of AI and the ecological and societal injustice propping this market up.
On a seemingly random day in March, Swedish pop star Zara Larsson posted a photo, appearing to be in bed, captioned: “Sorry I can’t hang out today, I gotta see what’s happening with chocolatina and strawberto.” The backlash was swift, with fans criticising that post and her reposting of videos from the AI-generated TikTok series, Fruit Love Island.
Some fans came to her support and defence, commenting that others were overreacting or that they could not bring themselves to care about Larsson’s consumption of AI content that she herself did not partake in making. The situation may perhaps appear to some as yet another episode in the ineffective and endless slog of online cancel culture. And yet, Larsson’s controversy is, through no meaningful intention of her own, a sign of our political times.
The environmental detriment of AI is no secret. Yet academics, corporations, and proponents of AI have extensively and consistently listed out the potential benefits of the technology in recent years, with LSE’s Dr Ruhi Khan even suggesting ways in which AI can be useful to the Global South. This glosses over a truth that is incredibly inconvenient to many: the Global North’s casual enjoyment comes alongside the ‘random’ suffering of the Global South that is actually connected to Larsson and many others’ ‘normal’ behaviour.
While it may be easy to accept that AI is bad for the environment, as the occasional cheeseburger is bad for the heart, you and you alone are both responsible and the victim of the consequences that eating the cheeseburger may have. The same cannot be said for using generative AI. While it’s environmental impact is known, precise numbers and visualisations make it harder to blissfully ignore.
According to Shaolei Ren, an associate professor at the University of California, Riverside, the water systems used to cool the equipment in the servers through which ChatGPT prompts run “keep [the servers] functioning”, which requires a great deal of electricity. More specifically, a single 100-word email generated by ChatGPT “requires 0.14 kilowatt-hours (kWh) of electricity, equal to powering 14 LED light bulbs for 1 hour”. That same 100-word email requires 519ml of water, which is just above the amount of water in the disposable bottles we buy at the shops.
This makes data centres (where servers are housed), “often among the heaviest users of water in the towns where they are located…driving up residents’ power bills and taxing the electric grid,” according to environmental advocates. It is therefore unsurprising that people living next to data centres have reported being unable to drink their water or live in their homes.
And of course, that is not AI’s only problematic aspect. Mega tech corporations outsource the “work” of AI to human workers in the Global South, who work in “inhumane” conditions for virtually no pay. Additionally, the minerals in microchips powering AI servers are “often mined in environmentally destructive ways.” These are a mere fraction of the perils AI causes and poses (that we know of), none of which are yet felt by the majority of the Global North. And yet, a silent but quickly growing minority exists and suffers daily, because Zara Larsson needed to know what’s happening with Chocolatina and Strawberto.
It would be ridiculous to pretend the blame falls squarely on Larsson or a teenage boy in Kentucky who are entertained by AI slop. It is, however, similarly ridiculous to pretend that engaging with and reposting generative AI, whether framed as fruity Love Island-esque “content” or a seemingly harmless and casual recipe search on ChatGPT do not contribute to the harm suffered by a disadvantaged many.
Casual is the key word here. Everything about Larsson’s post, fans’ reaction, and response may seem like arbitrary bad luck for Larsson, with a random moment blown widely out of proportion. And whilst people argue with each other and leave her comments online, millions of people will still use ChatGPT, and plenty of people in the rest of the world will continue to suffer and probably die because of AI. Indeed, how casual it all is.
To read more Comment analysis, please click here.
About the author
