The MediSafe Scandal Isn't About One Student. It's About a System.

Let me start with the technical facts, because the public conversation has skipped over them almost entirely.

An AI medical application of the kind described in the MediSafe award entry would require, at minimum: a working understanding of machine learning model architecture or the ability to integrate and fine-tune a pre-existing model via API; medical domain knowledge sufficient to define what the model is classifying or recommending; a frontend interface; backend logic; some form of testing or validation against real medical data; and documentation coherent enough to present to judges. Building something that deserves to win a major innovation award — as opposed to a functional prototype that demonstrates concepts — adds several layers on top of that.

I'm a computing teacher. I teach ICT at secondary level. My senior students are bright, motivated, and some of them are genuinely gifted programmers.

None of them could build this alone in Form 4.

That's not a criticism of any particular student. It's a description of what the work actually involves. The prerequisite knowledge — probability theory, data preprocessing, model evaluation, medical regulatory context — is not Form 4 content. It's not Form 6 content. It's undergraduate-to-postgraduate content. A Form 4 student building a sophisticated AI medical app unaided would be as remarkable as a Form 4 student publishing a peer-reviewed medical journal article. Not impossible, theoretically. But extraordinary enough to demand serious scrutiny.

The internet applied that scrutiny. The awards were returned in August.

What I want to talk about instead

The technical implausibility is, in some ways, the least interesting part of this story. Yes, the project was almost certainly ghost-built. Yes, someone — a professional, a parent's contact, a hired consultant — likely did the core work. Yes, this is cheating. Yes, it's wrong.

But here's my actual question: why does the ghost-building industry for student competition projects exist at all?

It exists because we built a system that pays.

In Hong Kong's educational ecosystem, innovation competition awards are not just trophies. They are credentials. They appear in Direct Subsidy Scheme applications. They appear in university portfolios. They are mentioned in school newsletters, parents' WhatsApp groups, LinkedIn profiles of parents who "supported" their child's achievement. An award like the one MediSafe received carries tangible downstream value in the competition for limited school places and university spots.

When something has that much value, and when the work required to earn it legitimately is genuinely beyond the reach of the target age group, you have created a market. And markets fill.

The ghost-building industry

I want to be clear: the MediSafe case is not unusual. What's unusual is that it became public.

In fifteen years of teaching in Hong Kong, I have seen competition projects submitted by students that were, by any honest assessment, not built by those students. The tell isn't always obvious from the outside — sometimes a project is just implausibly polished, implausibly documented, implausibly complete for its stated development timeline. But among teachers, there's a kind of quiet recognition that certain projects arrive bearing fingerprints that don't belong to fifteen-year-olds.

We mostly don't say anything. The reasons for that silence are worth examining — but that's a different article.

The industry that has grown up around this demand is sophisticated. There are consultants who will, for a fee, design a project architecture, write the code, prepare the presentation materials, and coach the student on how to answer judges' questions. Some of them advertise openly. Most operate through referrals in the enrichment centre and private tutoring networks where Hong Kong's education industry really runs.

The parents who use these services are not, in most cases, bad people. They are people who have accepted the logic of the system — that credentials matter, that competition wins matter, that the gap between what a child can genuinely do and what they need to show on paper must be managed. The system taught them this. The system rewards it.

What this costs the child

This is where I want to be direct, because it tends to get lost in the outrage cycle.

When a student submits work they didn't do, they miss learning. That sounds obvious. But think about what that means concretely in the case of an AI engineering project.

The Form 4 student who submitted MediSafe did not spend months working through the genuine difficulties of building AI software: the failed experiments, the debugging sessions, the conceptual restructuring when an approach doesn't work, the frustration of data that doesn't behave as expected. They didn't develop the resilience that comes from solving hard problems. They didn't build the mental models that come from understanding why something works.

What they got instead was a credential and a story they can't fully own. That's a worse outcome, educationally, than submitting a genuinely mediocre project and learning from the experience.

I have watched students who won competitions on the strength of their own work — projects that were rougher, less polished, genuinely Form 4 in scope and execution — go on to study computer science at university and build real things. The credential was less impressive. The education was real.

The systemic question

The question I keep coming back to is this: how did we design an education system in which the credentialling mechanisms are so decoupled from actual learning that ghost-building a project is rational?

Innovation competitions are meant to identify and reward genuine student ability and effort. When they instead become competitions to see whose parents have the best professional networks, they have failed at their own stated purpose. The awards went to the wrong students. The signal is corrupted.

DSE, for all its limitations, has one important structural advantage: it's assessed under conditions the student can't outsource. The exam hall is, in a very real sense, the truth-telling mechanism of the system. Everything before it — the competition wins, the portfolio, the STEM projects — exists in a different accountability environment.

I'm not saying competition projects should be abandoned. I'm saying the conditions under which they're evaluated need to account for the reality that many of them are not what they claim to be. Judges who interview students seriously — who ask them to explain design decisions, debug live, articulate their understanding of the technical choices — would catch ghost-building quickly. Most judging panels don't do this rigorously.

The MediSafe scandal is not the story of a bad student. It's the story of a system that incentivised, enabled, and rewarded performance over learning — until, for once, someone noticed.