Andrew Varsanyi

Vibe Coding and the Humanities: I Told You So

A while back, I wrote a post arguing that the AI-pocalypse everyone keeps predicting isn’t going to destroy the humanities — it might actually save them. I called it “The Coming Humanities Gap.” The core argument was simple: AI needs good prompts more than it needs anything else, and the people best positioned to write those prompts are the ones trained to read carefully, think critically, and write clearly. I thought I was right then. The rise of vibe coding is starting to prove it.

For those who haven’t encountered the term yet: vibe coding is the practice of building software through natural language. Instead of writing syntax, you describe what you want — in plain English, with as much specificity and detail as you can muster — and an AI model translates those instructions into working code. The quality of what these models produce has improved dramatically over the last several months. Anthropic, OpenAI, and Google have all shipped new models that are genuinely better at this, and the gap between “a developer wrote this” and “an AI wrote this” is closing fast. We’ve all read stories like this one about tech firms replacing engineers with AI.

Here’s the thing about vibe coding that no one in tech seems to want to say out loud: it is almost uniquely suited to people with humanities training. Writing a clear, useful, effective prompt for an LLM is not a technical skill — it’s a language skill. It rewards specificity, precision, and the ability to anticipate ambiguity before it becomes a bug. Those are not skills that emerge from a computer science degree. They emerge from years of close reading, writing, and argument.

I spent a good part of my career in tech before returning to graduate school, and one pattern I saw repeatedly was that the best development managers and project leads often didn’t come from technical backgrounds. They came from history, English, philosophy, political science. They succeeded because they could read carefully and think critically — and that let them navigate the enormous volume of information a software project generates. Design documents, technical specs, user stories, QA reports: all of it requires the same fundamental skill set that a humanities education builds.

That world is about to become the new norm, not the exception. Technical software knowledge is being commodified at a pace that should alarm anyone who spent four years learning to code and nothing else. Writing and running a vibe coding session, by contrast — knowing what to ask for, how to describe it, how to evaluate the output, and how to iterate — is much harder to automate. It requires genuine human judgment. It won’t be long before “prompt writer” and “prompt architect” become serious job titles, and the people filling those roles will disproportionately come from the humanities.

Historians — and I’ll admit my bias here — are probably the best positioned of all. Historical training is, at its core, about taking a large and messy body of evidence, making sense of it, and translating that sense into clear, precise prose for a reader who wasn’t there. That is almost exactly what good vibe coding requires. You are, in effect, briefing an AI on what you need, synthesizing your goals into language it can use, and then evaluating the result against your original intent. I can think of few more directly applicable skill sets for the AI moment we’re living in.

The honest caveat is that humanists who want to thrive in this environment will need to close a real knowledge gap. You don’t need to write code, but you do need to understand the landscape well enough to ask intelligent questions. How does a database work? What does it mean for an application to store state in the browser versus on a server? What’s the difference between a front-end and a back-end problem? These aren’t deep technical questions, but they matter. The humanities student who can’t answer them will be limited in what they can build, regardless of how good their prose is.

Which is why universities need to get moving. Digital humanities courses — real ones, not token offerings buried in the course catalog — should be required, not optional, for undergraduates and graduate students alike. The goal isn’t to turn historians into developers. It’s to close the gap between humanistic thinking and technical context just enough that these students can walk into a room with engineers and hold their own. We have a genuine opportunity here: to bring humanities sensibilities — nuance, skepticism, respect for intellectual effort, attention to evidence, care for language — into technology firms from the inside. Vibe coding is the door. Universities just need to help their students walk through it.