The Privilege of Agency
AI Is Not Replacing You. It Is Exposing You
A few years ago, we confidently told an entire generation to learn coding. Software engineers were the aristocracy of the Information era, since they pocketed the trick of propagating information at scale. They possessed a rare bridge, the ability to translate human intention into machine-readable instruction. That bridge created leverage, status, and insulation. If you could speak to machines, you were safe.
In a recent interview, Jensen Huang made a simple but unsettling observation. For years, we believed software engineers were among the most talented and privileged professionals of the modern era. And now, look at what AI is first targeting, the act of writing code itself. The irony is structural. The class we assumed was the safest in the digital economy is also one of the first being abstracted by the tools they helped build. Not eliminated overnight, but compressed, assisted, automated, and redefined. That doesn't diminish engineers. It reveals something deeper: no skill remains sacred once it becomes patternizable. And AI is a pattern-recognition engine at scale.
When I look at this moment, I'm less interested in the fate of engineers and more interested in what remains invariant across centuries. Civilization has never been advanced by syntax. It has been advanced by psychological posture. The people who build new things are rarely just the most technically fluent in the outgoing system. They are the ones who sense that the system itself is shifting and adjust early. They are curious enough to experiment before consensus forms. And they possess the agency to act while others debate.
Ridicule is always part of the pattern. Today, I see three reactions to AI. The first group ignores it, behaving as if it's a loud but temporary trend. "We've seen hype cycles before," they say. "This will pass." The second group treats it like a fetish. They try ChatGPT once, generate something mildly clever, and conclude the experiment prematurely. The third group quietly adopts it — and often gets looked down upon for doing so.
There is a strange moral undertone in some circles: if you use AI for writing, coding, research, or thinking support, you are somehow cutting corners. As if leverage were dishonorable. As if calculators destroyed mathematics or spreadsheets ruined finance. But tools always look like shortcuts before they become literacy. Dismissing AI fluency today is not a sign of depth; it is often a sign of inertia.
For decades, machines could not understand the languages we speak. They required translation into rigid syntax. Software engineers became powerful because they were bilingual, fluent in both human ambiguity and machine precision. Code was the intermediary layer between thought and execution. Now that intermediary is dissolving. Large language models increasingly understand ordinary language. English itself has become an interface. You describe intent; the system translates it into executable instructions. The abstraction layer has moved upward.
The literacy frontier has shifted. Fluency is no longer about memorizing syntax. It is about articulating intent clearly, structuring problems well enough that a system can act on them, and iterating toward better outputs. The new leverage is not in typing faster. It is in thinking sharper.
This is where agency becomes decisive. AI makes execution cheaper. When execution becomes cheap, direction becomes scarce. When direction becomes scarce, clarity of thought compounds. If your thinking is fuzzy, AI outputs are fuzzy. If your prompts are shallow, results are shallow. AI does not rescue poor judgment; it amplifies it. It rewards those who can define what is worth building and why.
History has a familiar rhythm. The printing press threatened scribes. Electricity frightened cities. The internet was mocked as a fad. Early adopters often looked slightly foolish, until infrastructure caught up. The people who move forward are rarely those who perfectly mastered the outgoing paradigm. They are those who were willing to look slightly naive while learning the incoming one.
I don't pretend to forecast exactly how labor markets will rearrange themselves. But I am comfortable making one narrower claim. When the cost of building collapses, advantage shifts to those who can decide what is worth building. That has never been a purely technical trait. It has been a character trait, the ability to ask uncomfortable questions, the curiosity to explore without immediate validation, and the agency to pursue an idea even when the room is skeptical.
AI democratizes competence. It does not democratize conviction. And when competence becomes abundant, conviction becomes rare.
If I were advising a young person today, I wouldn't tell them to avoid AI in order to preserve intellectual purity. I would tell them to become fluent with it, seriously fluent. Not as a toy, not as a shortcut, but as leverage. Learn how to extract work from it. Learn how to structure problems for it. Learn how to sharpen your own reasoning through it. Because the real divide will not be between engineers and non-engineers. It will be between those who wait for the future to stabilize and those who participate in shaping it.
Agency was always the multiplier. AI simply removes the excuse of friction.