Proletarianisation: What We Delegate to Machines

The question we dare not ask

The scene is a pleasant one, this time. The tool has been deployed, it works: the minutes write themselves, summaries drop in a few seconds, everyone has gained several hours a week. The steering committee speaks of an “adoption success.” And somewhere in the room a question hovers, one no one asks aloud because it would sound ungrateful: what if I were losing something?

This question is wedged between two discourses that keep it from existing. On one side, the injunction to save time: delegate, optimize, don't fall behind — to worry is to have failed to understand. On the other, a diffuse guilt that dares not argue its case: the vague sense of cheating, of going soft, without being able to say precisely what would be lost or why it would matter.

And yet you already know the answer, through an experience almost everyone has had. Ten years of GPS, and the day the phone dies, the city you cross every day turns unreadable. No one stole your sense of direction: it dissolved, gently, into the delegation. You lost nothing the day you installed the app — you lost it with every day you no longer exercised it.

The previous episode closed on a question left open: what exactly do we lose when we delegate? To answer it, Bernard Stiegler built his most incisive concept.

Three waves of dispossession

The proletarian is not who you think

The word will come as a surprise: proletarianisation. It smacks of the nineteenth century, of class struggle, of manifestos. But Stiegler takes it back from Marx and shifts it: the proletarian, on this reading, is not defined by poverty — he is defined by the dispossession of his knowledge. The artisan mastered an entire process; the factory worker pulls levers whose logic he no longer understands. His gesture has not disappeared: it has been captured — discretized, recorded, incorporated into the machine. Readers of the previous episode will recognize the mechanism: it is a grammatisation. What is grammatised becomes delegable; what is delegated without return ceases to belong to anyone.

First wave, then: manual savoir-faire — that of gesture and hand, the first to be captured.

Desire under GPS

The second wave, in the twentieth century, climbs higher: it reaches savoir-vivre. Marketing and the cultural industries grammatise modes of existence — what we eat, what we watch, what we desire. The ideal consumer no longer knows how to live on his own: his existential choices are prescribed, his tastes are segments.

If this description sounds dated, its contemporary version is not: the attention economy is exactly that — a GPS for desire. The recommendation feed decides the next video, autoplay decides that there will be one, the notification decides the moment. Each of these micro-delegations is comfortable; their sum is a savoir-vivre exercised less and less. You recognize the signature of proletarianisation by this detail: it is never a theft — it is an erosion by consent, one service rendered after another.

The wave that strikes the powerful first

The third wave is the one that concerns us directly, and Stiegler dates it to a precise event: the 2008 financial crisis. What it revealed goes beyond finance — some of the best-paid decision-makers in the world no longer understood the algorithmic systems trading in their name. Savoir-penser itself — to analyze, to judge, to decide — had dissolved into automation. A decisive detail: this wave does not strike the weak first. It strikes first those who have the means to delegate everything.

Generative AI extends this movement to everyday reasoning: drafting, summarizing, comparing, deciding. And here is where we must state the criterion that saves this concept from catastrophism. A delegation does not proletarianise in itself. It proletarianises when the delegated knowledge ceases to be exercised and understood — when no one, nowhere, grows in capability. Delegating calculation to the calculator did not proletarianise engineers: they understand what it does and could redo it slowly. The criterion is not the delegation; it is what remains alive in the one who delegates.

What shifts and what dissolves

Now we must face the best objection, for it is a solid one. Every generation mourns lost knowledge — Socrates already mourned the living memory that writing would weaken — and yet humanity has never known as many things as it does today. Knowledge does not disappear, this objection says: it shifts. We lost mental arithmetic and gained programming; lost our bearings and gained the time to think about something else. The balance would be positive, and proletarianisation a learned name for nostalgia.

The objection deserves to be taken seriously, and Stiegler contradicts it only halfway: he does not deny the displacement — he asks whom it benefits. When writing captured memory, entire institutions (the school, the law, libraries) organized the return of knowledge to people: the displacement produced literate minds. Proletarianisation names the inverse case: the one where knowledge leaves people without returning anywhere — where it dissolves into the system. 2008 remains its purest demonstration: the knowledge the traders lost, no one gained. Not even the machines, which executed without understanding. So the question is never “are we losing something?” — we always lose something. The question is: is someone, somewhere, growing in knowledge thanks to what I delegate?

Brought back to concrete work, this question becomes immediately operative. The automated minutes: if no one rereads them as the person accountable, the knowledge of synthesis has dissolved. The HR pre-screening tool: if no one can justify a rejection other than by invoking the score, the knowledge of judgment has dissolved. Clinical decision support: if the practitioner can no longer say why he agrees with the machine, something of the craft has dissolved — even if each individual case is better handled.

I have lived both sides of this process. For fifteen years I built software whose function, in the final analysis, was to automate tasks — and so to displace knowledge — at my clients' firms. Then I made the reverse journey: retraining in psychology, that is, spending years slowly re-internalizing a knowledge that cannot be delegated, that cannot be installed, that can only be exercised in the first person. I can testify to it plainly: deproletarianisation is not an abstract concept. It is work — long, costly, and one of the most invigorating things I have ever done.

The upkeep of knowledge

The inventory of delegations

From this episode, take away a tool that fits in three questions — to ask once a year, on your own or as a team:

1. What have I delegated this year that I no longer ever exercise? The list is longer than one thinks; writing it down is often enough to set priorities.
2. For each delegation: do I still understand what the machine does in my place? To understand does not mean being able to redo it quickly — it means being able to explain, to contest, to spot the error.
3. Where am I growing in knowledge thanks to the time saved? This is the question the injunction to save time never asks. If the answer is “nowhere,” the time was not saved: it was displaced — and probably captured by something else.

The counter-movement exists

Proletarianisation is not a fate, and the counter-movement is not a utopia: it has proof. Wikipedia is written by people who grow in knowledge by contributing. Free software trains its own maintainers. Stiegler and the Ars Industrialis group called this the economy of contribution: devices where use itself produces knowledge in the user, instead of extracting it.

At the scale of a team, deproletarianisation looks like modest things: automated minutes that someone rereads and signs; a junior to whom one explains what the tool has just done and why it is almost right; regular stretches of practice without assistance — not out of purism, but the way one keeps up a muscle one hopes never to call on in an emergency. Relearning a route by heart is not nostalgia: it is upkeep.

Driving with map and GPS: that, at bottom, is the position this series has defended since the first episode. Neither refusing the tool nor dissolving into it — staying capable of both gestures, and choosing, in full awareness, which one to exercise today.

Knowledge is not a stock

If this episode had to fit in a single sentence: knowledge is not a stock we possess, it is an exercise we keep up — and what we cease to exercise, we first lend, then lose.

One question remains that the three waves leave intact: where, exactly, does all this captured knowledge go? Into which supports, into whose hands, and with what power over our memories? Stiegler has a word for these supports — and it bears directly on the language models that memorize in our place. Next episode: tertiary retention, or our externalized memories.