Why Generative AI Fluency Does Not Settle Next Month

Introduction

How to use generative AI, and how to fold it into your working day, are already well documented. What this article takes up is what comes after that. Once you are fluent, what accumulates?

Endless material exists on becoming proficient, and this one point gets skipped: what proficiency actually changes about your economic position. Skip it, and you arrive back at the same conclusion every time — I am still not using it well enough.

Here is the conclusion first. A skill is labour, not capital. However advanced the craft, on its own it cannot leave the structure in which income arises only while you are moving. The question is not old skills against new ones, nor how proficient you are. It is who holds the ground on which value is produced.

Fluent, and Still Uneasy About Next Month

People who use generative AI daily carry a strange sensation. The work is measurably faster. The range of what they can do has widened. And yet the unease about next month’s income has not lifted in proportion.

The dissonance does not come from a shortfall in ability. Getting faster and becoming stable are different problems.

Raise your working speed and you can handle more in the same hours. But the source of income has not moved. Someone commissions you, you deliver, you are paid. Stop moving and the income stops with you. Raise the speed without touching that structure, and the root of the unease stays where it was.

What Actually Set the Price Was Not Your Craft

Why has a high price ever been paid for one competence and almost nothing for another? Stated in a phrase: asymmetry of knowledge.

Information asymmetry describes a transaction in which one party holds more information than the other. The economist George Akerlof formalised the idea using the used-car market [Akerlof, 1970, Quarterly Journal of Economics] — where the buyer cannot judge quality, the viability of the market itself is affected.

Skilled work rested on the same structure. Writing well, producing a clear visual, translating accurately — these take years to acquire, and someone who has not acquired them cannot even judge why the result is good. That inability to assess is precisely what the fee was paid for.

Credentials, apprenticeships, long training periods: each of these maintained the asymmetry institutionally. Physicians and lawyers are paid well not only because the work is difficult, but because the number of people holding that knowledge is limited by design.

From this an uncomfortable implication follows. The value of a skill sat not inside you but in the distance between you and the market. As more people can do the same thing, whatever you gained by improving is absorbed by the market.

What AI Compresses Is Not Your Ability

Generative AI compresses that asymmetry along two routes.

• Democratisation of production — competences that took years to acquire are shortened by assistance. People without the competence can produce output of a usable standard. The moment only some people can do this becomes anyone can, scarcity disappears
• The logic of substitution — from the commissioning side, if the quality gap falls inside tolerance, choosing the cheaper option is rational

What matters is that this compression is not happening because your ability is low. High or low, the asymmetry itself is shrinking. It is a structural change in the market, not a problem of individual capability.

And as the asymmetry thins, the ground of competition shifts. A market that ran on who is best becomes a market that runs on who is cheapest and who is already known. When technique levels out, everything other than technique decides the outcome.

“Skill at Using AI” Is Compressed by the Same Force

Then acquire skill at using AI — a natural response. But this is nothing other than repeating the same problem one storey higher.

Proficiency with these tools is scarce at a given moment. That scarcity is compressed too, as the models improve and as more people become proficient. The trajectory prompt engineering travelled in a short span was the preview.

The chain has no end. Acquisition cost falls → more people hold it → scarcity falls → price falls. No amount of individual sharpening runs against that chain. Hunting for the next scarce competence is an endless chase between the speed of the technology and the speed of human learning.

This is examined in one thing to check before learning prompt engineering.

Five Adaptations, Two Dependencies

The strategies attempted to escape this cul-de-sac come to roughly five: updating your skills, hyper-specialising, raising output efficiency, building a product, differentiating through a personal brand. They point in visibly different directions.

Abstract them, and every one falls into one of two categories.

• Skill dependency — investment in raising the quality or quantity of your ability. Whatever changes about the type or level, the heteronomous structure in which it functions only when someone happens to need it does not change
• Platform dependency — placing your business on someone else’s foundation. That side alters rules and fees as its own business requires. You remain permanently on the responding end

The two reinforce each other. Selling a competence requires a marketplace to reach buyers; leaning on that marketplace produces platform dependency. Avoiding platform dependency leaves skill dependency intact and merely converts the marketplace into direct outreach. Moving between the two, choosing which is currently less bad — that is the enclosure built by the question how do I adapt to AI?

None of these strategies is wrong. Each is a rational answer. What determines the shape of the answer is the shape of the question.

Capital Is What Detaches From the Body

Here the language needs to be exact. Capital, in this article, means a non-linear structure in which one unit of labour keeps producing N units of output across time.

The core is non-linear. Skilled labour is linear: work an hour, get an hour of result. But something written once continues to be read after it was written. A mechanism designed once functions while its designer rests.

Gary Becker’s human capital theory positioned investment in skill as capital formation [Becker, 1964]. That such investment raises productive capacity is a fact. But the theory demonstrates that skill raises productive capacity — not that skill produces a 1:N structure of output. A highly skilled person can command a high hourly rate. If that person does not move, the income is zero.

There is a further, decisive property. Capital detaches from the body of its owner; skill does not. A factory runs while its owner is ill. Writing is read while its author sleeps. Your competence, by contrast, produces value only while you are in motion.

That inseparability is the core reason why skill, however advanced, does not become capital.

History Has Already Shown This Once

This structure is not new.

Before industrialisation, skilled artisans held both the craft and the tools. Holding both gave them a degree of autonomy. After the machines arrived, their craft did not deteriorate. They kept the craft and lost the means by which it could be exercised. Which left working in someone else’s factory as the remaining option.

Karl Marx analysed this change along the axis of who owns the means of production [Marx, 1867]. Those who hold them produce goods and take the return. Those who do not sell their labour power to those who do.

The economist Kozo Uno systematised this as double freedom [Uno, 1964]. The first freedom is freedom from bonds of status. The second is the freedom of holding no means of production. A person with both looks free. They can change jobs. They can go independent. But the structure — using someone else’s means of production, selling labour to someone else — does not open.

This is the answer to “why did nothing get easier after I went independent?” The first freedom changed; the second did not. Formal independence and a change in economic structure are entirely separate questions.

The history is taken up in the industrial revolution took tools, not skill and what capitalism’s double freedom actually means.

The Same Technology Works in Opposite Directions

Here is the asymmetry at the centre of this article.

For those competing on flow, generative AI is a headwind: it increases supply and pushes rates down. For those holding stock, it is a tailwind: it accelerates accumulation. The same technology, working in opposite directions depending on position.

The same fork existed during industrialisation. Some smashed the machines. Others bought a small machine, set up a workshop and functioned as small-scale owners. What decided the difference was not their attitude to the technology but their position with respect to the means of production.

And something is now open to individuals for the first time in history. Publishing writing once required a publisher’s approval; releasing music required a contract. For content, that gatekeeper is no longer required. What to send, to whom, and when, can be decided by you.

This is treated in leverage is not a multiplier borrowed from finance.

What Counts as a Means of Production Now

So what is there to own? Not a factory and not machinery. A structure through which you can decide who receives what you make. Concretely, three things.

1. An audience — a direct relationship with readers that does not pass through an intermediary, is not tied to an account, and does not vanish with a change of terms
2. Intellectual assets — accumulated work embodying a worldview, where something written three years ago is still read today
3. An automated system — a mechanism that keeps delivering without your direct involvement

The third is treated in automating the work does not give you the path; the design of the delivery path itself in why blogging longer leaves less of your own voice.

Two Things Said Plainly

First, this is not an argument against developing skill. Skill matters in two ways. Deeper expertise produces deeper insight. Stronger expression turns first-hand material into something that lands. And skill is the basis of trust. The question here is sequence and priority. Sharpening a competence without a structure to deliver it, and sharpening it while holding one, produce entirely different results from identical effort.

Second, building takes time. Several months to more than a year, during which results are hard to see. AI lowers the cost of that period; it cannot remove the period. Trust accumulates over time, and while the rate of accumulation can be raised, the time cannot be skipped. Describing an automatic mechanism while omitting this would be dishonest.

Conclusion: From How to Adapt to What to Own

Fluency with generative AI does not settle next month because the shortfall was never proficiency. However advanced a competence is, it cannot be detached from your body — and what cannot be detached does not function as capital.

And while the question remains how do I adapt to AI, the answer lands on skill dependency or platform dependency. The shape of the question has already fixed the shape of the answer.

The exit is not greater precision in adapting but replacing the question with: how do I own a structure that keeps functioning whether or not AI arrives? On that single point, the same technology begins to work the other way.

The economic side of this argument is developed across the economics of structural autonomy.

Articles in This Cluster: Reading Each Argument in Depth

What is happening, and where it hurts

• Why crowdsourcing rates fall, and it is not your craft — not skill, not negotiation: what happens when the source of value disappears
• Check this before debating when the singularity lands — not the date, but who is distributing the question

Taking the adaptations apart, one at a time

• One thing to check before learning prompt engineering — a new competence rides the same cycle of obsolescence
• Who gets the time your efficiency gains free up? — speed does not produce ownership
• Whose foundation is what you built with no-code on? — being able to build and being able to hold are different

Rereading it through history and structure

• The industrial revolution took tools, not skill — artisans kept the craft and lost the position
• What capitalism’s double freedom actually means — free of status, and free of the means of production

Moving to the side that owns

• Leverage is not a multiplier borrowed from finance — how many times does something made once arrive?
• Automating the work does not give you the path — automating tasks and owning arrival are separate

On adjacent ground: how far independence actually resolves is taken up in going independent does not reduce the number of masters; owning the path of delivery in why blogging longer leaves less of your own voice; the entry point for those still employed in why a side hustle only adds another buyer for your hours; and where the standard for money sits in why earning more does not reduce the anxiety.

The book-length treatment of this argument is available as Your Skills Are Not Your Capital on Amazon Kindle.

References

Academic papers and theory

• Akerlof, G. A. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism” (1970) Quarterly Journal of Economics, 84(3)
• Becker, G. S. Human Capital: A Theoretical and Empirical Analysis (1964) Columbia University Press

Books

• Marx, K. Das Kapital, Band I (1867)
• Uno, K. Principles of Political Economy (1964) Iwanami Shoten
• Rifkin, J. The Zero Marginal Cost Society (2014) Palgrave Macmillan