Reading about AI can be taxing. A refrain of so many people I know, from computer science professors to journalists to bleeding-edge social scientists, is that they cannot keep up with how much there is to read about AI. Everyone complains about this, even when you know your lane and stick to it. There is a palpable anxiety that the flood of writing prevents you not just from not reading, but from not being sure that how you navigate what to read within the limited time there is to read anything is rigorous or effective. If you try to read everything then you will actually never end up reading anything because you have not figured out the most efficient way to read what needs to be read. I’m being facetious about this kind of optimisation, of course. Which is why I came up with a German word for this and settled on Diskursüberflutungsmüdigkeit. It means: ‘fatigue from the flood of discourse’. (I used deepl dot com, my imagination, and a passing acquaintance with the German language).

I read selectively about AI. When I am not reading academic books and articles for work, then I am on the lookout for writing by academics and critics that bring together a close reading of technology with an assessment of the sociopolitical, cultural, economic or geopolitical stakes of the technology. I usually want something current and topical because academic publishing cycles are longer. Yet how can something ‘current’ also be thoughtful and reflective; these things take time to write. Or should they?

So I scan a few Substack pieces a week, usually in a desultory fashion, as if I am being forced to. This is the part of my reading optimisation that is unfocused and flailing; I think it is because a number of computer scientists, political scientists, and economists working on AI write on Substack, and they write fast, and I want to read them all but obviously cannot.

But here are some of the readings (and one ‘listening’) that have brought joy these past few weeks and months.

Quinn Slobodian, Digital Bandung in the Ideas Newsletter.

Slobodian is a historian and in this essay he stress-tests the rise of the words empire and colonialism as suffixes to the words tech, data, and AI, to ask if these are accurate at all, and what the implications are of treating our current moment in terms of of colonial relations. He writes "‘Actual scholarship teaches us that using the word “empire” as a straightforward synonym for a systematic, centralized, and overwhelming form of rule can be precisely wrong.’ And he goes on to tell us why, that empire and colonialism words are statecraft words that don’t map on to the full experience of digital and datafied life. Stans of Couldry and Meijas ‘data colonialism’ will bristle here, but he’s right. In many an academic seminar (including one with Professor Couldry) I’ve heard a lot of caveat-ing of what data colonialism means, and how the two authors use the word ‘colonialism’. I struggle with the language of coloniality and empires with respect to tech and AI, because many usages tend to reproduce a narrow conceptions of periphery/core, and ignore the distinct ways datafication, power, extractivism, and territorialisation emerge. I get it though, naming Big Tech as ‘empire’ has offered a way to name power and in doing so, award shame and seek accountability if not actual reparation.

Slobodian offers, as a counter, three terms that he believes should also be used to discuss datafication: ‘the comprador, the creole, and counter-colonization.’ . He writes:

Many authors endorse a dynamic in which the plundering colonizer extracts fortunes at the barrel of a gun and leaves nothing but poverty and death in its wake. But empire itself did not usually work in this clear-cut way. It rather operated through a middleman called a comprador class.

He proposes that we swap political categories for economic ones, and talk about resistance in the same moment as oppression. People working and writing right in the middle of all this will find the examples and citations familiar. This novel approach highlighting the dominance of a particular set of ideas is really refreshing; and he give us new vocabulary that help us expand how we frame our fights, analysis, critique.

(I can talk for a long time about the impoverishment of language in some contemporary movements; but that’s another topic.)

While I’m on Slobodian, I also highly recommend Muskism: A Guide for the Perplexed (with Ben Tarnoff). It’s not what I expected and I’m really glad I read it. It is only to an extent about Musk himself, and it is actually about how the sociopolitical and economic environment has changed and made it possible for Musk to become who he is now. (He was responsible for changing a lot of it; in that sense, he’s a remarkable futurist).

Meghan O’Gieblyn, ‘We did our best!’ in the New York Review of Books.

I will read anything and everything by O’Gieblyn; she used to be a nun, and an alcoholic, and left the church and addiction. I believe this makes her a reflective and scholarly person who writes about technology and philosophy. Her book God Human Animal Machine: Technology, Metaphor, and the Search for Meaning (Anchor Books, 2022)is one of those touchstones for going back and reading about Western philsophical traditions and Christian influences on understandings of mind, spirit, and the body, and their relationship to how the West understands tech/AI. It is also a great primer to philosophies that you may not have formally studied. (And that I have not studied).

In this text she draws out the parenting metaphors circulating among technologists and in Silicon Valley culture who believe they are responsible for ‘giving’ human values to agentic AI systems. I understand how compelling the parenting metaphor can be; it is metonymous with a kind of care that we want to show and share but do not always have the socio-affective or economic infrastructures to see through. Hence the coy but insipid ‘plant parent’. I mean, I use it myself but not without wincing. Is parenting the only language we have for caring for someone or something? Anyway, back to AI. A recent documentary about AI’s existential risk, The AI Doc, similarly has men folding their concern for the planet and the human species with the birth of their children. O’Gieblyn exposes how from philosophy to computing this metaphor of parenting AI remains a powerful way to think about ensuring accountability and values in technology; however, she notes, the same people also actively disregard or resist institutional regulation that does the very same thing.

Marek Polliks and Roberto Alonso Trillo, the Disintegrator podcast: The datacenter does not exist.

The third thing is not actually a piece of writing, but a piece of listening. I’ve been recommending The podcast was recorded as part of a lecture tour for their book, Exocapitalism, about how capitalism can now operat without being tethered to the materialiality of labour and resources. This episode is about how the data centre exists not just to generate electricity, but to generate value by funnelling contracts, packages of debt, and speculative risk financing through the economy. The data centre is therefore a financial instrument of some power. Hence, the episode argues that data centres not being built would not make a difference to the people who actually stand to gain, which is not just tech companies and includes markets and banks. After listening to that episode I felt like something fell into place for me in understanding a dimension of the debate beyond data centres’ environmental impact.

I hope you enjoy these! Till next time (which I hope is sooner rather than later)

Maya

And please remember to subscribe!