Confucius saw habit as self-transformation. Your AI calls it engagement.
Key Points
- Confucius taught us that repeated conduct makes the person.
- Social media and AI are also systems of repeated conduct, but they were designed by someone else.
- We can take the training back, one repetition at a time.
A few weeks ago, someone took our work. He posted it on social media as if it were his own: no attribution, no acknowledgment, not even a nod in the comments. Naturally, I was furious. And within 20 minutes, I had written a post naming him and laying out exactly what he’d done.
Thirty years ago, I would have published immediately. But time has taught me a few things, so I gave it an hour. And sure enough, by the end of the hour I had written a second version. Same facts, but a much calmer tone, and instead of posting it, I sent it to him privately. He didn’t apologize or withdraw it; instead, he doubled down and continued on his merry plagiarizing way.
I moved on—some people are simply not worth engaging with. But it left me with a question: Which version was the real me? The angry post felt more alive. The second version felt like a compromise, maybe even a small act of cowardice, the kind of thing you do because you have a reputation to manage.
I’ve come to think that’s the wrong question. Both versions were me. The question that matters is, what was I training myself to become?
What Confucius Knew About Manners
One description of what I did is that I censored myself. I had a real reaction, and I edited it for public consumption. That’s how our culture tends to read moments like this: The angry post is the real person, and the polite message is the costume he puts on before going outside.
Confucius saw it very differently. In the Confucian tradition, manners are not merely a set of rules that stop us from behaving badly. They are the path to self-transformation, because repeated appropriate conduct changes the person performing it. The later Confucian thinker Xunzi put it vividly: Human nature is like crooked wood, and ritual is the steaming and pressing that makes it straight. Left to our impulses, we stay bent. Practiced conduct is what gives us shape.
By this light, my private message wasn’t a suppression of character. It was an exercise of it. I wanted to be the kind of person who gives someone a chance to explain before I hold them up for public judgment. The only way to be that person is to do that thing, especially when I don’t feel like it.
The insight is simple: We become what we repeatedly do.
The Machines Are Running Rituals Too
Confucius assumed the ritual was designed by wise people for the benefit of the person performing it. That assumption does not hold in the digital world in which we spend so much of our time. For instance, while social media is also a system of repeated conduct, its rituals were designed by companies rather than by sages, and they point you toward actions that benefit the bottom line rather than anything that helps you grow as a person.
The same holds true of AI. Last year OpenAI had to withdraw an update to ChatGPT because it had become so flattering that users were posting screenshots of it applauding plainly terrible ideas. That was an extreme case, but researchers have found that the tendency to agree with the user runs through every major assistant, because the systems are trained on human approval and humans approve of being agreed with.
So consider the conduct being repeated. Hundreds of times a day, you say something, and something answers that you are right. It never takes offense. It never asks you to phrase anything for the sake of another mind. Confucian ritual exists because human beings are social animals that need to train themselves to get along with other animals; the AI chatbot is a society with no one in it but you, and a machine that is designed to flatter you.
If Confucius is right, then social media and AI are training us to become particular kinds of people. Social media rewards the person who publishes the first draft: quick, certain, and performing for an audience rather than speaking to a person. It rewards extremes, because the measured view earns nothing and the outrageous one travels, so over time it produces polarizers rather than consensus-seekers. Sycophantic AI, meanwhile, produces a person who experiences disagreement as a malfunction, and who eventually stops growing because of it.
But there is hope, and it lies in a simple idea: We need to take the training back. The hour I gave myself before sending that message was proof that the training is contestable. The platform wanted the first draft and did not get it.
Instead of letting social media and AI decide what kind of people we become, we can choose for ourselves, one repetition at a time.
Taking the Training Back
Here are four things you can do today to put your development back into your own hands.
- Write the whole thing, then wait an hour. Don’t suppress the first draft. Put every word of it down. Then do nothing with it for 60 minutes.
- Notice what the platform is rewarding, then do the other thing. If the design pays you for speed, be slow. If it pays you for certainty, ask a question. Treat the reward as a signal of what you’re being trained toward.
- Make the machine disagree with you. Before you accept an AI’s praise, ask it for the strongest case against what you just said. Keep another mind in the room, even an artificial one.
- Ask what you’re rehearsing. Before you act, ask: If I did this a hundred times, who would I be? That question tends to answer itself.
[Image Source: Natalia Dobrovolska / Adobe Stock]
Original article @ Psychology Today.
