In the AI Era, the College Degree Is Becoming a Blank Sheet of Paper

Some recent thoughts, shared here for discussion.

1. This year, no one seems to be talking about the gaokao

Around this time in past years, my feeds and group chats were full of the gaokao — good luck, score estimates, picking universities. This year I noticed it clearly: much quieter.

At first I thought I was imagining it. I wasn’t. More than “how to get in,” more and more people are starting to ask a different question: once you’re in, is it still worth it?

I think that shift matters a lot. It’s not that people are less anxious — it’s that they dimly sense the ticket they’re fighting so hard to grab is losing its value.

My own read is blunter: in the AI era, the college degree is becoming a blank sheet of paper.

2. A university really only sells two things

To explain why, you first have to look at what a university actually sells. Peel it back and there are only two things:

The signaling value of the diploma — a piece of paper that helps an employer quickly judge “roughly what tier of person you are” before they’ve even met you.
The teaching of standardized skills — a body of knowledge, procedures, and routines, loaded into your head in batches, by semester, by syllabus.

For the past few decades, these two things were very valuable. The diploma was the door knocker; the skills were the rice bowl. But here’s the problem — these are exactly the two things AI flattens first, and most completely.

3. The first thing: the diploma’s signal is dropping to zero

Why is a diploma valuable? Because it solves an information gap: the employer doesn’t know you and can’t vet everyone one by one, so they use “what school you went to, what you majored in” as a rough sieve.

But the thing AI is best at is eliminating information gaps.

When an employer can have AI look directly at what you’ve actually done — your projects, your code, your work, what clients say about you — do they still need that piece of paper to “guess” whether you’re any good? No. When something can be verified directly, the signal stops being worth anything.

This isn’t hypothetical. You can already see the early signs in the last couple of years: some of the most cutting-edge tech companies are starting to hire high schoolers and people who never graduated, with a simple rationale — “Put you on a real project for four months and I’ll know if you’re good. I don’t need the diploma at all.”

The diploma used to be a passport. Once everyone can be “read” directly by AI, it reverts to what it always was underneath: a blank sheet of paper.

4. The second thing: standardized skills are losing value too

Someone will say: fine, the diploma’s useless, but the real skills I picked up at university are still mine, right?

Which brings me to the point I think matters most: our education has long disguised “the ability to solve problems on a test” as “the ability to solve problems.”

What is an exam? The questions have been written for you, the conditions laid out for you, the standard answer already exists, and your job is to find it by the shortest path within a time limit. That’s a powerful skill — Chinese students are trained to be extraordinarily strong at memory, calculation, and performing under pressure.

But the real world isn’t like that. The real world has no given conditions, no standard answer, and at the start often no one even knows what the problem actually is — a product isn’t selling: is it the pricing, the channel, the wrong demand, or the wrong direction entirely?

The hardest thing in the real world has never been “solving”; it’s “posing and defining the problem.” And that is exactly what school trains you for least.

Worse still: that “follow the steps, execute the standard moves correctly” ability is precisely the part AI now does best and cheapest. A while ago I used AI to do my taxes for the first time — it asked the questions, did the downloads, ran the checks. Something that used to mean days of back-and-forth with an accountant got sorted out in minutes. The gap between someone who’s never studied finance and someone with a master’s in it is shrinking fast in front of AI.

When “being able to do it” becomes nearly free, spending many years learning “how to execute the standard moves correctly” yields a pitifully low return.

5. Being clear about what I’m not saying

At this point I want to be especially clear, to avoid being misread:

I am not saying learning is useless. What’s breaking down is the specific path of “trading a diploma plus standardized skills for a white-collar job” — not all learning and effort.
I am also not saying the role of school is useless — especially when children are young, in primary and secondary school, what school really means is something I haven’t figured out for myself. That’s a more complex topic for another day.

What I want to say today is just one thing: the university, and the “diploma signal” and “standardized skills” it sells, are depreciating fast in the AI era. The more I think about it, the more certain I am.

A note at the end

If both the diploma and standardized skills are depreciating, then what should we actually accumulate in the AI era?

I’ve written my answer before — it’s reputation, the real, verifiable traces you leave in this world. A diploma is a piece of paper someone hands you; reputation is something you earn bit by bit, that no one can take away.

The difference between those two may be the single thing most worth figuring out in the next ten years.

These are my personal thoughts — discussion and pushback welcome.