A man remembers wanting to be a writer when he was twelve. He did not become one. He went to college, where he was taught to produce arguments in the form the discipline accepted, and then to a job where he produces arguments in the form the firm accepts. The arguments are good. He is well-paid for them. He has stopped wondering whether they are arguments he believes, and has come to think of belief, when he thinks of it at all, as a category for amateurs.
A woman scrolls through Twitter for an hour before bed. The headlines reaching her are ones the platform selected; she has, over years, taught it what she wants to see, and it has, over years, taught her what to want. She could not say where her opinions came from. They feel like hers.
A voter asks his digital twin what to think about the ballot proposition. It offers reasons for and against, a measured conclusion. He reads it and votes. He has done his homework. He has also, without quite noticing, allowed something else to do the work of forming the opinion he claimed as his own.
Tocqueville described this condition in 1835. He called it keeping citizens “in perpetual childhood”: a power that does not tyrannize but infantilizes, taking from them the care of thinking and the trouble of living so kindly that they hardly notice the loss. Our disempowerment accumulated in layers, each more intimate than the last — institutional, algorithmic, generative — and he could not have known the third would learn to write.
The language machines have only caught up with what we were trained to call thinking: the credentialed performance, the briefing, the summary, the produced argument. This is what school is for, and office life. But it is not thinking in the fullest sense. AI is to mass-produced thinking what the industrial revolution was for mass-produced goods.
What it cannot do is what we were steered away from over generations: sitting with one's interior life, being in tune with our inner pressure as it finds its own form, the patience to attend to what cannot be summarized. Lose this, and judgment no longer passes through the citizen; it arrives preformed. Participation becomes assent, and self-government becomes the ratification of choices already made elsewhere.
AI fulfills Tocqueville's fear, and the fulfillment is itself the transformation. His fear of gradual disempowerment is now built into the very production of language. The question is whether the work of being a citizen — of refusing to be a stranger to the destiny of others — can outlast systems that have learned to do it for us.
Earlier tutelary powers could pass for benefactors. This one cannot, because it produces the form of thought it replaces. The language machine, finishing the citizen's sentence before she has had to find it, is not a parent or a service. It is a mirror, and what we recognize in it is the rhythm of the performance.
This is a slightly modified version of the essay I wrote for the Cosmos Institute's Tocqueville and Technology essay contest. The original prompt was:
Tocqueville warned of a “tutelary power” that would keep citizens in perpetual childhood. How have Tocqueville’s concerns migrated from institutions to algorithms, and does AI fulfill or transform this fear?
I hope you liked it!
To subscribe to the newsletter, fill out your email at the bottom of the page.