When AI Can Do Almost Anything, What Will Set Students Apart?

I went to high school in Melbourne under the VCE system. In Year 11, I completed Maths Methods ahead of schedule and received a perfect score, placing me roughly in the top 0.3 percent of students across the state.

When the results came out, everyone around me said the same thing: “You should tutor.”

It was the obvious choice. People were already willing to pay an hourly rate that felt substantial to a high school student. I could turn the score into one tutoring hour after another, take on a few students each week, and earn very good money for someone who was sixteen or seventeen.

I did not do that.

I opened my laptop and started making a pitch deck. I wanted to turn the way I had learned mathematics into a complete course.

I am telling this story now not because starting a company in high school is something every student should copy, but because AI is changing what was actually valuable about the experience. Today, a student working with AI might build in days a system that took me much longer, while also producing early course content, product designs, and marketing materials more quickly. As execution gets cheaper, what sets people apart is no longer only what they made. It is who found the problem, who chose the direction, and who kept moving once reality answered back. That is the real reason this company kept coming up in my university and job interviews.

One choice nobody assigned

I moved from China to Australia at thirteen. My English was poor when I arrived, and I often could not follow lessons. Mathematics was one of the few things that did not depend as heavily on English and that I could still hold on to. I spent a great deal of time on it and gradually found methods that genuinely worked. Completing Maths Methods early and earning a perfect score meant more to me than a number on a transcript. It meant I had found something that might be useful to more people.

But between having an idea and making it real stood a long list of things I did not know how to do.

I was sixteen. I had no business plan and no clear idea how any of this should work. I made the pitch deck first, trying to explain the idea well enough for someone else to understand it. Then I spoke to people wherever I could and gradually pulled together a team of six. I did not really know how to lead people, or even what an education product was supposed to include. I often committed first and then rushed to learn how to deliver.

I redesigned the curriculum around the path by which I had actually learned mathematics rather than copying the order of the school textbook. Then came recording. I went to school during the day and filmed at night. A fifteen-minute lesson often took two or three hours to record. Altogether, I spent hundreds of hours recording and editing, usually alone at my computer. Then there were notes to write, exercises to create, and learning materials to organize.

When the course was ready, I assumed the hardest part was over. Then I realized students could not simply watch a collection of videos. They needed accounts, progress tracking, and a way to return to the right place. So I began building my own learning management system. It was the first real system I had built from scratch.

Then a more practical problem appeared: the course existed, but nobody knew about it.

I started doing sales and marketing. For a sixteen-year-old, it was naturally uncomfortable. I had to explain the course to strangers, respond to parents' questions, and deal with hesitation and rejection. Once the first students joined, their feedback was direct. A concept I thought I had explained clearly still confused them. A feature I had spent a long time building was barely used.

So I changed it. Then changed it again.

The project gradually grew. The course eventually served more than three hundred students and began producing steady passive income. I have always had mixed feelings about those numbers. Three hundred students is not a huge scale, and the revenue was not extraordinary. What mattered was that the product had entered the real world. People were genuinely using it to learn, and when it fell short, they said so. It was no longer an idea on my computer. It was something for which I was responsible to other people.

Interviewers were not interested in the word “Founder”

When I applied to university, this became the most important extracurricular activity in my application. Not because “high school founder” was an impressive label, but because the experience contained so many facts that could be examined: a six-person team, hundreds of hours spent recording and editing, an LMS I had built myself, more than three hundred students, and real revenue. I did not need to write that I had leadership or that I cared about education.

The difference was even clearer in interviews. Interviewers rarely asked about abstract qualities. They asked why I had built an LMS instead of using an existing tool. Where had the first students come from? Which feature had failed? What would I change if I started again?

Those questions have something in common. If you have truly done the work, you can keep answering. If the project was mostly packaging, you run out of substance quickly.

At UC Berkeley, I studied computer science and mathematics. When I first applied for opportunities at leading technology companies as a freshman, I had almost no conventional software internship experience. Compared with older students, my résumé looked thin. But once I brought up what I had built in high school, interviewers listened and kept following the details. By my sophomore year, the project was still one of the experiences they returned to most often.

It could not replace algorithms, programming, or systems knowledge, and it could not guarantee the outcome of any interview. But it gave interviewers a reason to evaluate me seriously. They could see that I had already dealt with real users, a real system, and responsibility for a team, not only coursework. My judgment also grew through everything I later studied and did in university, but that high school project was where I first began to practice it.

Looking back, the deepest mark it left was not a line on my résumé. It let me experience a complete loop very early: find a problem, organize the resources, make a first version, put it in front of people, receive feedback, revise it, and ship again. I still run that loop today while building Lantr and the AI company Predexon.

When making something gets easier

Students today face a completely different environment.

AI can help write code, create designs, assemble prototypes, organize research, analyze feedback, and produce content. Young people can now build, at much lower cost and much earlier, things that once required a team. The cost of execution is falling quickly.

That creates another change that is easy to miss.

As making something becomes easier than it used to be, a polished output becomes less persuasive on its own. A website, an app, or a research report can now appear very quickly. Universities and companies will find it increasingly difficult to judge someone by the finished work alone. They will care more about what sits behind it: Who found the problem? Which decisions did the student make? Were there real users? What changed after the feedback arrived? Who owned the result?

AI can analyze a problem, offer options, and even surface something you have missed. But the choice of direction, and responsibility for its consequences, cannot be handed to a tool. What is scarce in the AI era is not simply the ability to use AI. It is the ability to use AI to turn a problem you discovered into a real result, while keeping direction, judgment, and responsibility in your own hands.

Do not turn projects into another scheduled class

When families think about AI, the first response is often to add another class: coding, prompting, an AI competition. But if adults still supply the question, the steps, and the answer, the student may simply be using a new tool to complete an old kind of assignment. The tool changed. The person making the decisions did not.

What a student may need is not another scheduled activity, but a project nobody assigned. It can be small: investigate a question, build a community tool, design materials for a school, write a report someone can actually use, or start a modest service. It does not need to be a company, produce income, or even succeed. But it must begin with the student, meet real people, and receive real feedback.

There are only a few things parents need to do, but they matter.

First, do not define the entire project for the student. If adults choose the topic, steps, team, and schedule, the student is still completing an assignment. Agency requires that the choice remain theirs. Parents can help narrow the scope, connect resources, and protect sensible safety boundaries, but they cannot take over the central question and decisions.

Second, find a real user before worrying about application value. That user might be a classmate, a teacher, a club, or a small local business. Scale matters less than honest feedback. A small tool someone genuinely uses says more than a polished project nobody opens.

Third, do not look only at how polished the result appears. Ask what judgments the student actually made. Why this direction? Where did AI help? Which suggestion did they reject? What was wrong with the first version? What did users say? Who decided the next step?

A grade shows that a student can complete a task someone else defined. A real project shows what they do when nobody tells them the next step.

AI will give more and more people the ability to turn ideas into things. What sets students apart may not be who learned a particular tool first, but who will still choose a problem when nobody assigns one, put a first version in front of the real world, and take responsibility for what comes next.

That first project gave me a simple kind of certainty: I did not know whether it would work, but when nobody gave me the next step, I knew I could begin.