The Glass Hammer and the Human Touch
In a world where AI writes poems, composes symphonies, and generates lines of code faster than we can blink, it’s tempting to imagine a future where programmers are simply no longer needed. But that future, despite the hype, is far from arriving. Inspired by the thoughtful work of Ted Chiang, whose writings on artificial intelligence (notably in The New Yorker) dissect the assumptions behind our tech myths, this piece explores why the human programmer still matters—and always will.
Chiang argues that AI is less like a mind and more like a compression algorithm—a powerful, fast, and sometimes confounding tool, but not an intelligent being. In his essay “Will AI Become the New McKinsey?” he writes: “Think of ChatGPT not as a mind but as a blurry JPEG of all the text on the web.” That image is as poetic as it is precise: what AI offers is a hazy, approximate reproduction of existing patterns, not genuine understanding or intention.
This distinction matters. Coding is not just typing; it’s thinking. Tools like Cursor, Claude Code, and Gemini CLI can autocomplete functions or scaffold applications, but they do it from patterns, not context. They don’t understand your product vision, the nuance in a user’s experience, or the trade-offs that surface in a planning meeting. They are brilliant assistants, not creators—magnifying glasses, not architects.
And yet it would be a mistake to ignore them. AI is no longer optional in the day-to-day practice of software development. From autocompleting boilerplate to generating tests, summarizing PRs, or untangling spaghetti logic, it has woven itself into our workflows. Programmers who refuse to adapt won’t be replaced by AI—but they may be replaced by other programmers who have learned to use it well.
This is the distinction often lost in alarmist headlines: AI isn’t replacing developers, it’s reshaping development. It makes good programmers faster and fast programmers better. Those who embrace the shift spend less time on tedious work and more on architecture, collaboration, and the actual problem. Those who resist find themselves outpaced—not by machines, but by peers.
In this landscape the human touch becomes more important, not less. Understanding how people behave, thinking critically about data, working with non-technical teammates, exercising judgment about what should be built: these are the places AI isn’t equipped to shine. If anything, its rise raises the value of distinctly human work.
We should stop picturing the programmer as a keyboard monkey easily swapped for a language model, and start recognizing what programming actually is: as much art as science. It’s not about writing loops—it’s about solving problems. And AI, for all its capability, doesn’t know what the problem is until a human defines it.
Chiang’s framing is a useful lens: don’t treat AI as a synthetic mind, but as a mirror—one that reflects our own assumptions and possibilities back at us. It isn’t here to dethrone us. It’s here to challenge us.
The future doesn’t belong to AI. It belongs to the programmers who understand how to wield it.
That conviction shapes how we build. If you want people who think this way building your software, let’s talk.