I used to build artificial intelligence. For years, I worked on systems that understood what people said, answered their questions, and did their work for them. Unexpectedly, that work led me back to a question about people. The more I saw what AI could do well, the clearer I became about the one thing it could never do. As it took on more of our work, I began to ask what it means for a person to experience something for themselves.
The advantage has shifted from technical skill to taste
AI has changed the terms of the game. It has dramatically lowered the level of technical skill needed to make something real.
Technical expertise used to give us a defensible advantage: knowing a field, writing code, composing an essay, drawing a picture. These skills defined our professional identity and helped determine our value. Now AI performs much of that work, and much faster. The ability to produce things is no longer what protects our place. When almost anyone can make something, knowing how to make it no longer makes us exceptional.
What, then, distinguishes us? Knowing how to frame a problem. Recognizing what is good, beautiful, or compelling. As the ability to make things becomes common, the judgment to decide what deserves to be made becomes more valuable. Taste is the new advantage. Not a preference borrowed from everyone else, but taste developed through experiences we have lived and savored ourselves.
Why taste? Because questions fall into two broad kinds: those with a fixed answer, and those whose answers depend on what people value. What is the solution to this equation? Where there is one correct answer, AI has already outperformed people, and it should. But what is beautiful? What is right? What is good for me? There is no answer key for these questions. Human subjectivity determines the answer. Yesterday's right answer can look dated today; something once dismissed can come to seem right. No formula settles the matter.
This is where AI hesitates. It is designed to choose answers that people are likely to accept: safe answers, broadly agreeable answers. Faced with a question that has no fixed solution, it tends toward the acceptable average. It accommodates its audience and gravitates toward whatever the most people will approve. But deciding for ourselves what is good is a responsibility AI cannot assume for us. Nor should we let it. Once we hand that decision over, we are no longer directing our own lives.
Let me clear up a possible misunderstanding. I am not saying AI cannot understand taste. It can generate perfectly decent choices in an instant. That is precisely the problem. When decent taste is everywhere, it loses its distinguishing value. What remains is the final, decisive difference: the one part that turns nine out of ten into ten. The part AI cannot reach can only come from what someone has lived through personally. A conversation with a friend. The memory of coming apart. A sunset that moved us one evening. These are not things a system can acquire by learning about them. They belong to the person who experienced them.
Taste grows in people who savor their lives
Taste does not simply come to everyone.
Taste grows out of a personal sense of beauty. If an aesthetic is our own way of recognizing what is beautiful, taste is that judgment expressed in what attracts us and what we choose. It develops in people who have discovered, through their own experience, what moves them, rather than simply following what others admire. To cultivate taste is to develop a personal aesthetic. And that grows only by living and savoring experiences, one by one, rather than consuming them and moving on.
The question I arrived at while building AI brought me to savoring experiences so they become part of a life: giving a passing moment our own attention and thought, then considering what we found good in it. Over time, these moments help us recognize what moves us. No one, and no AI, can live those moments for us. We develop our own grounds for choosing by experiencing and savoring things ourselves.
Capitalism has long been driven by a sense of lack. We try to avoid looking inadequate. We act because we feel short of something, fear falling behind, or believe we must acquire more. That feeling makes us grasp for things, pushes our judgment outside ourselves, and eventually leaves us powerless. But there is another source of motivation: an appreciation of beauty. Recognizing something beautiful and feeling enriched by it can make us agents in our own lives. And someone who can act for themselves can also take responsibility. I wanted to write about a way of seeing the world that is driven by beauty rather than inadequacy. The point is not to abolish the feeling of lack. It is to cultivate a different motivation alongside it, gradually shifting from being pushed by inadequacy to being drawn by beauty.
The question of remaining irreplaceable in the age of AI eventually becomes a question about what we consider a good life. It asks us to make choices from a sense of beauty developed through our own experience, rather than simply adopting someone else's standards.
The experiences that shape what we choose
One autumn evening a few years ago, I stopped during a walk in a park to watch a golden sunset. I probably watched for about three minutes. Yet the light kept returning to me for days, at home, the next morning, and afterward. The moment was brief, but because I stayed with it and savored it, it remained with me.
AI can make a great deal for us, but our sense of what is good takes shape through experiences of our own. A cup of tea, a conversation, a brilliant sunset: moments we have lived and savored help us decide what matters. Choosing a life according to our own taste begins with giving those experiences our attention.
