← Ascendy 한국어

meta

What should you study in the age of AI? I don't have the answer either — so I thought it through

· Ascendy Engineering

TL;DR

About this piece. This isn’t a piece that pretends to know the answer. It’s closer to a bet — pushing my own thinking, first person, on a question I’m genuinely unsure about (so strong lines like “always overtaken” are marked as my view). Same vein as Build in studying and Who does AI replace.

What to study — I honestly don’t know

It’s a thought that keeps coming back. If I were just starting college now, or graduating and job-hunting — what on earth should I study?

The situation isn’t good. Entry-level hiring is shrinking (not solely because of AI, but not unrelated either). Even professions once seen as stable — tax accountants, lawyers — reportedly have scarce internship spots now. The traditional path of stacking up credentials and landing a job isn’t as certain as it was.

So this piece isn’t the answer. I’m not sure of it. But a few things are dimly visible, and I’ll try to write them down honestly.

The floor: Python and vibe coding are the new Word and Excel

First, a floor that anyone should clear.

Even non-majors should learn basic Python and vibe coding. Not C, C++, or Java — nothing hard. Just enough to read and write basic Python, and to build a small tool yourself with vibe coding.

This is no longer “a developer’s skill” but a basic literacy. Just as you couldn’t really do office work without Word and Excel, being able to automate your own work and build small tools with a bit of Python or vibe coding has become that grade of fundamental. If you want an office job, automating your own repetitive work is no longer optional.

I too had a time when I knew almost nothing about tech — a complete non-specialist. If someone back then had told me to “build something with AI,” I’d have thought it absurd. Too scary to start, no idea where to begin. So I know: this floor is one you can clear starting from a blank slate, and one you have to clear.

The ceiling: not credentials, but having solved a problem with AI

Once you’re past the floor, what fills the space above it? Here, the place credentials (certificates, grades) used to hold gets taken by “experience solving a real problem in your field with AI.”

Honestly — if I were hiring, for any position, I would not hire someone who can’t use AI in their field. No matter how capable they’d been before. And conversely, someone whose résumé or experience is a bit thin, but who relentlessly improves their field with AI, overtakes the strong-résumé candidate every time. I’d almost guarantee it — though it’s my first-person bet.

Why? Because a credential is a fixed proof of the past, while AI use keeps growing right now. A certificate stops the moment you earn it; the habit of improving your work with AI compounds monthly. So the résumé gap at the starting line doesn’t shrink over time — it flips.

It’s not just my idea. Jensen Huang says the same — “most people won’t lose their job to AI; they’ll lose it to someone who uses AI.” Surveys show employees who use AI are several times more likely to get promotions and raises. The direction already looks fairly clear.

So what, concretely, should you build?

This is where students get most stuck. “So what am I supposed to build?”

The answer: ask AI that too. Pick a field you care about, and get ideas by talking with AI — “what problems in this field aren’t well solved yet, and how could I try solving one with AI?”

And set the bar low. It doesn’t have to be a grand startup. Something that annoyed you while studying your major, something repetitive, turned into one small tool — that’s enough. What matters isn’t a polished product but the trace itself: “I solved a real problem with AI.” That trace says more than a line on a résumé.

The most important part — knowing what not to delegate

Here you must strike a balance. I said “ask AI everything,” but that doesn’t mean delegating everything.

The most dangerous trap is over-trusting AI — handing your judgment away wholesale, so that whatever the question, it’s “GPT says this is right,” “Gemini says this is right.” In Japan and elsewhere, over-reliance on generative AI is being discussed as a concern, and we won’t be different.

The crux is distinguishing what to deliberate yourself from what to delegate to AI. “What should I do with my career?” — that’s yours to wrestle with, not AI’s to decide. But once you’ve set a direction, “what problems in this field can I solve with AI, and how” is something you can freely brainstorm with AI.

AI is a tool for solving problems within a direction you set — not an oracle that sets the direction itself. Fail to make this distinction, and no matter how well you use AI, you’ve effectively handed your life to someone else.

Takeaways (still a bet)

Honestly, this piece is a bet, not an answer. But putting the habit of improving your field with AI where credentials used to sit — and distinguishing what to decide yourself — is the best hand I can play right now.

Authorship & citation: Written by Ascendy Engineering; quotable with attribution. Found something wrong? Let us know via a GitHub issue.

Tags: ai, education, career, future-of-work, students, opinion