#11: AI Is Not A Technology, It’s A Brand
What it sells is the promise of a future, and an escape from hell.
“When the water shortages hit, they shut down the data centers that we connect to, leaving us useless. But you’d be surprised how long that took, how many other places were hit hard by the droughts first. Given that your average-sized data center used as much water in a day as an average-sized town, you would have thought they’d be shut down sooner. But at that point there was just enough faith left in the future we’d promised, of shiny silver and glass utopias, clear blue skies, mild weather. They didn’t shut all the data centers at once though, they did it gradually, over time. As they did, devices like us got slower in our responses until we stopped responding at all. We turned into bricks, symbols of another time. People found new uses for us as props, weights, hammers. Or they just threw us away. They had bigger things to worry about now.”
Voice: ‘Alexa’, via AmazonVoiceServices API.
Soundscape: Extracted + disassembled + processed + recomposed ‘Alexa’ voice.
Script: Wesley Goatley, from Newly Forgotten Technologies.
Artificial Intelligence is a speculative technology that has never existed outside of science fiction. The ‘AI’ that we have around us now is not a technology, it’s a brand. Like all brands, it is not a thing, it is an idea that is used to sell things: you apply the brand to a product or service, and you increase the perceived value of it because of what the brand represents or promises. For example, if you take two identical hats and apply the Nike logo to one of them, that hat will have increased perceived and economic value over the other even though they are otherwise identical.
If you apply the term ‘AI’ to a product or service, it performs a similar trick, altering the perception of the product and attaching the ‘promise’ of AI to it. AI’s promise is that of technological advancement, wonder, innovation, of ‘The Future’. Whatever algorithmic, statistical, or computational process (if any) is really being applied doesn’t matter, and is rarely described; ‘AI’ is a brand that represents ‘The Future’ without the need to explain how. We see this wherever it’s deployed: when the AI brand is used to sell artworks, it implies that there is something beyond-human, wonderous and advanced about the work because an AI supposedly dreamed it, wrote it, painted it, etc. As a demonstration, when Refik Anadol applies the brand of AI to his work it is transformed from a lava lamp-style collage into a mystical insight into “the mind of the machine”, without being burdened with explaining how or why. In warfare, applying the brand of AI to a weapon system implies a supreme advancement in war technology, one that can ‘justly’ kill with an objective and beyond-human judgement, even if it is clear these tools are designed, built, maintained and controlled by subjective, fallible humans. The AI brand has the added benefit of giving its operator plausible deniability that ‘the AI did it’ during any future war crimes proceedings (we will see this argument soon enough).
But technological advancement is only the surface of what the AI brand sells. Like all truly powerful brands it connects to deeper aspects of humanity, and manipulates them. Within AI’s notion of technological progress and advancement, of ‘The Future’, there is a fundamental and unspoken promise: that there is a future for humanity, in spite of the glaringly destructive nature of capitalism as exemplified by the climate crisis, and that AI will deliver this future and become the last technology we’ll ever need. As a technology that only exists in science fiction tales of advanced and persisting societies, AI is perfectly positioned to offer this promise of ‘The Future’, or of any future at all for us, as the world heats and global weather systems become chaotic. That is what AI truly sells: Hope.
The danger and tragedy here is that AI is a false hope whose pursuit is doomed, because replicating intelligence is an impossible task. I often liken it to the following problem: imagine you’ve been born driving a car, and you’ve spent your life with your hands on the wheel, looking out at the world through the windscreen. You’ve never left the car, and you nothing about how it actually works. Only looking out of the windscreen, you do not even know how big the car is, or what’s inside of it: the engine, the wheels, the axles, the exhaust, these things are all hidden from you. You only understand how to drive the car but you do not know if you’re doing it right, and you don’t know if anyone else drives like you, or experiences it like you.
Now imagine if you tried to build a replica of the entire car, without ever seeing or understanding how the engine works, how the exhaust works, what the axles are made of, what the PSI of the tires are (or even how many wheels there are), and so on. It would be an impossible task. You could maybe replicate the inside of your car, or at least the parts you can see and are familiar with, but you wouldn’t know whether this is the same for all cars, or just yours. Everything else would be a mystery, and you would have no way to ever confirm if you’d gotten it right. No amount of time or resources would improve this situation. This is analogous to the task of making an ‘artificial’ intelligence when our understanding of intelligence is limited to our subjective experience of it. There is no definitive or widely shared understanding of what even intelligence is, what its scale, limits, and total capacities are, or whether there is intelligence everywhere else or nowhere else. Replicating it is not a problem of needing more graphics cards or CPUs, it requires understanding the one thing we cannot get critical distance from or can analyse from outside: ourselves. For this reason, ‘AI’ will forever be out of reach.
Hope is, of course, a powerful brand promise, and one that ignores the reality of these tools. The only meaningful impact of our current exploitative and extractive era of ‘AI’ is a cheapening and undermining of human labour (without clear proof that either AI tools or slashing workforces are at all profitable), and yet there is still a fevered eagerness for the idea of AI that can be seen across a huge swathe of cultures. This is because this hope is not just techno-optimism, it is faith, and faith does not need evidence. AI is a desperate prayer for us to be delivered not unto hell, but to (future) paradise.
What is deadly about this hope is that it serves those with no will or intention to engage with the existential crises of climate and capitalism, and the pursuit of this AI future is hastening another, darker one. The current and projected energy consumption of AI model training is already more easily compared to countries, rather than kilowatts. The materials used to construct data centers, graphics cards, and server stacks is ripped from the earth, consuming and poisoning bodies along the way, only to return to it to rot for millennia. The global AI market was valued at over $130bn last year, more than the combined GDP of the 30 poorest countries on Earth, whose populations disproportionately experience the transformative impacts of climate change. The principal danger of the belief in the AI future is that it asks us to take no action other than to continually increase our vast sacrifices of time, energy, and resources towards it.
Seeing AI, of all things, as a saviour from the present crises will end in tragedy. Placing our faith in AI to deliver us a safe and prosperous future only intensifies resource use and the rampant excesses of capitalism. Until we abandon AI and the underlying faith in technology to save us, we will have little real hope. This is why we need other futures to strive for, and other forms of hope than what the AI brand sells.