“Alexa, what is the real cost of your switching on my lights?”

Here’s a link to a stunning diagram, and the must-read accompanying long article, called “The Anatomy of an AI system.” I understand it’s won some kind of award for a design as iconic as, perhaps, the London Underground map of Harry Beck; or perhaps closer still, those diagrams of the cell’s biochemical processes that so impressed me with God’s wisdom during my medical training.

The difference from the latter is that, whereas the huge complexity of cell processes is all about a network of co-operation for the common good of all, the AI diagram, although it shows similarly complex interactions, also reveals multiple fractal pyramids of exploitation of many people for the good of the few.

I don’t want to detail too much here, and stop you reading the piece itself. I just want to whet your appetite by pointing out some of the surprising truths about AI (and thus about our whole computer-heavy lifestyle) that are hidden, perhaps carefully hidden, from us.

To us, as the article begins, an Amazon Echo appears a simple wonder of clean technology: the neat little drum called Alexa will, at our command, switch on the lights, play Michael Jackson tracks, or answer little Hermione’s questions about history. If there are any questions in our mind about the risks of AI, they tend to be at the theoretical level of future threats to our jobs or, even more speculatively, of AI taking over from humans altogether.

The reality is much more prosaic: the article (summarised in the diagram) documents how Alexa is actually the sweet-talking face of a global system that exploits a multitude of real people (including the purchaser, who in return for a few baubles of convenience also unwittingly becomes an employee and a material resource for the supplier). At one point those at the top of the pyramid, such as Amazon CEO Jeff, Bezos gaining $270 million a day contrasted with the 7 year olds working in the Cobalt mines in the Congo who would need to work 700,000 years non-stop to earn the same, are compared to the pharaohs of old, for whom all toiled.

We agonise on Facebook about banning plastic straws in the west (with a negligible effect on plastic waste, even without considering the resource and energy cost of the alternatives). But we are blissfully unaware of the minimally paid – and sometimes unpaid – indentured workers slaving away in Indonesia, or China, or South America to obtain rare earths for the processors, batteries and solar cells we are using to discuss it. Not only are their working conditions inhuman, but their mining and extraction industries are both fossil-fuel heavy and heavily polluting.

I heard yesterday that, on some estimates, we will use up the world’s supply of these rare elements in thirty years (just as we phase out the last non-renewable infrastructure). All we will have to show for it is a vast mountain of unrecyclable AI and batteries – the average life of a smart phone is just three and a half years, because everyone needs the latest and throws away the old.

But the energy use is as much in the operation as the manufacture:

Large-scale AI systems consume enormous amounts of energy. Yet the material details of those costs remain vague in the social imagination

As human agents, we are visible in almost every interaction with technological platforms. We are always being tracked, quantified, analyzed and commodified. But in contrast to user visibility, the precise details about the phases of birth, life and death of networked devices are obscured. With emerging devices like the Echo relying on a centralized AI infrastructure far from view, even more of the detail falls into the shadows.

Even when you logged in to read this, you were actually working, unpaid, for the pharaoh:

In a paradox that many of us have experienced, in order to prove that you are not artificial agent, you are forced to train Google’s image recognition AI system for free, by selecting multiple boxes that contain street numbers, or cars, or houses.

One aspect that surprised me was in a section built around the analogy of the “Mechanical Turk,” a chess-playing automaton that actually was worked by a concelaed grand-master within. It seems that much of the training of AI depends on labour intensive, and poorly paid, human workers abroad, whose mundane data input provides the intellectual legwork enabling the manicured “news anchor” Alexa to provide such easy answers to us. We see no any sign of the vast human labour that has saved us looking stuff up or reaching for the light switch:

As we see repeated throughout the system, contemporary forms of artificial intelligence are not so artificial after all. We can speak of the hard physical labor of mine workers, and the repetitive factory labor on the assembly line, of the cybernetic labor in distribution centers and the cognitive sweatshops full of outsourced programmers around the world, of the low paid crowdsourced labor of Mechanical Turk workers, or the unpaid immaterial work of users. At every level contemporary technology is deeply rooted in and running on the exploitation of human bodies.

Read the whole piece, please. There is a sense, of course, in which what it describes is the stuff of human life in any conceivable age: our race is a network of interactions between primary producers, those who add value, and those who consume the goods and services so produced. But in a well organised society – as in the well-ordered cell – there should be a high degree of repricocity throughout the system. I grow wheat and sell my surplus to you, and with the money perhaps I will buy one of the statues you carve, or one of the knives you forge.

But what strikes me most forcefully from this excellent piece is our complacency, as those somewhere closer to the top of the pyramid than the bottom. We don’t even ask, let alone show concern, about the evils inherent in the information technologies we take for granted. The convenience of our domestic AI is, when one considers it, a way of procuring the servitude of many others by proxy.

There were never slaves in England during the transatlantic trade. The ships left Bristol or Liverpool with trade goods for Africa, exchanged them for slaves in Subsaharan Africa, and sold them for cotton or sugar in America and the West Indies. The human cargos remained invisible.

Ordinary people in England got better clothing and dental caries; and even though we see the cotton mills as exploitative now, at the time they offered relatively well-paid employment opportunities. It is doubtful how much most people knew about slavery – and still less would they have appreciated its actual horrors until the abolition movement informed them in the nineteenth century.

Those who built the big houses we now pay to visit on the profits knew more about it – though perhaps their wives and children didn’t as they sipped their tea or tinkled a bell to have the lamps lit by servants they barely noticed. Yet in the end, Britain’s industrial economy was founded on chattel slavery, and we all say it would have been better if it had never happened.

A snobbish family friend of my mother returned from Apartheid South Africa, maybe fifty years ago, and blithely said (in my hearing), “They all talk about race problems in South Africa – but my dear, you hardly even see any black people.”

Out of sight, out of mind. Alexa, play “Pharaoh” by Richard Thompson.

Jon Garvey

About Jon Garvey

Training in medicine (which was my career), social psychology and theology. Interests in most things, but especially the science-faith interface. The rest of my time, though, is spent writing, playing and recording music.
This entry was posted in History, Politics and sociology, Science. Bookmark the permalink.

Leave a Reply