What It Means to Earn a Salary in the Age of AI

A public investigation notebook on what earning a salary means in the age of artificial intelligence. A thesis defended by a French payroll practitioner, tested against the real world through ongoing collection of concrete cases and an active search for contradictions.

Everywhere, people are talking about the disappearance of work.

I think we are addressing the wrong question.

What is disappearing is not work. It is the salary.

I have written a white paper on what it means to earn a salary in the age of AI. It defends a precise thesis: a salary does not pay for the execution of tasks. It pays for the human responsibility for what is produced. That is why AI, which cannot hold responsibility, cannot take a salary in the proper sense. It is humans who decide, or refuse, to keep paying other humans. AI is the argument. It is not the cause.

This thesis shifts the debate. It names a black hole of responsibility that no one is looking at. It introduces the question — political, proprietary — of who owns the tool that now produces the value.

But this thesis is not a conclusion. It is a position I will test against reality.

About the white paper

The white paper itself is currently only available in French. The original title is Qu'est-ce que ça veut dire toucher un salaire à l'ère de l'IA. An English translation is in preparation.

In the meantime, English-speaking readers, journalists, and researchers can:

What this site is

Not the final version of a book. A public investigation notebook.

You will find the white paper in its current version. You will also find, week after week, concrete cases collected from the field — freelancers whose rates are collapsing, employees whose roles are being transformed, companies replacing or keeping their humans, executives laying off, executives refusing to. And you will find the contradictions I actively look for against my own thesis. A thesis that has not been tested against its counter-examples has not been thought through.

As cases accumulate, the text will evolve. Some parts will be reinforced, others nuanced, others rewritten. Successive versions will be dated and preserved, so that one can see how the thinking has shifted — or solidified — through contact with the real.

This is the only honest way to defend a strong thesis. Not to plant it in the ground and defend it against every objection. To propose it as a working hypothesis, and put it to the test.

Why I can allow myself this

I have been producing payroll slips for over ten years. I teach others how to produce them. Every week, I see dozens of concrete cases pass through my hands — companies wondering about AI, employees wondering what is happening to them, executives hesitating.

This observation post entitles me to speak. Not as a theorist. As a practitioner who has learned to think from what he sees.

It also limits me. I see what I see — not the rest. I know the field of payroll in France and the French-speaking world — less so the Anglo-Saxon, Asian, or Latin American fields. My biases are those of a white European man working in an intellectual profession.

That is precisely why I publish in open mode. So that other perspectives complete mine. So that other terrains expand mine. So that the thesis, if it holds, holds not because it suits me — but because it has survived eyes other than my own.

What you can contribute

If you read this site, you probably know something I do not.

Maybe you are a freelancer whose rates have shifted since ChatGPT arrived. How exactly? On what kinds of assignments? With what clients?

Maybe you are an employee whose role has transformed. What did the machine take? What is left to you? What was added?

Maybe you are an executive who hesitated to lay people off invoking AI. What made you choose one path or the other? What did you discover afterwards?

Maybe you think my thesis is wrong. On what concrete case? With what arguments? I want to hear them — not to defend myself, but to think more accurately.

Maybe you are a lawyer, a sociologist, an economist who sees in this text an approximation that needs correction. Welcome. I prefer a correction that costs me to an approximation that suits me.

All these contributions will find their place here. With your name if you wish, under a pseudonym if you prefer, anonymized if necessary. I commit to handling them rigorously and to making clear when a case you transmitted has shifted my position.

What this site is not

Not a forum. Not a space where anyone publishes whatever they like. It is an edited investigation notebook — by me, under my responsibility.

Not a participatory consultation either. I will not change my thesis because a majority of contributions tells me it is wrong. I will move it if an argument, or a body of cases, shows me that it does not hold.

Finally, not a disguised commercial product. I will not sell you anything at the end of the journey. My professional activities exist elsewhere. This site is an intellectual project that asks for nothing more than to be read, contested, enriched.

The wager

I am betting that a strong thesis, defended honestly and exposed to contradiction, is worth more than a cautious thesis hiding behind precautions.

I am also betting that in the age of AI — when texts are mass-generated without any human answering for them — writing publicly under one's own name and accepting to be held responsible for what one writes becomes again an act that has value.

This white paper is not only a text on wages in the age of AI. It is also, by the way it is published, a test of its own thesis.

I sign. I answer. And I accept being wrong — if reality shows me so.

Now it is your turn — to look, to test, to contest if you wish.

If the thesis holds — really, after the cases, after the contradictions, after the ground — it will have deserved to hold.

If not, I will have moved it. And that will be another form of success. Perhaps the best one.

Anthony Roca

Payroll practitioner and trainer

Read the white paper in French →