Originally written in Traditional Chinese: 我們要的不是抓得更準，是讓孩子不必藏
A new report from the Center for Digital Thriving at the Harvard Graduate School of Education asks a question we rarely ask about AI in schools: what is it doing to trust?
In a spring 2025 survey, 74% of teachers and 69% of principals who described an AI dilemma named something related to cheating. That part is not surprising.
What stopped me was in the follow-up interviews with 31 teens, ages 15 to 19. Some students said they deliberately held back their best work, because they were afraid of being falsely accused of using AI. Principal investigator Beck Tench calls it "cognitive withholding," and notes it happens in environments where accusations are not followed by any real conversation.
I see a version of this all the time. In presentations, more and more colleagues point at their own slides and say, "Look, I really made this myself. There's a typo in it." In this era, a few mistakes and a little imperfection have become proof that a human made the work. Students write a little worse. Adults leave a typo in. Different ways of hiding, same fear: being mistaken for AI.
We usually worry that students will hand their thinking over to AI. This is the opposite problem: students hiding their thinking from us. And it is harder to spot, because it just looks like an ordinary assignment.
Policing AI works like a searchlight. The brighter it gets, the further students retreat into the dark, and their real ability goes with them. The goal is not better detection. It is making the thinking process visible, so no one needs to hide. As a school leader, this is the point I keep coming back to: the most important thing now is to leave a record of the process. That record is what lets anyone know, with confidence, that the work is truly yours. A typo proves nothing. The process does.
Two practical shifts follow from the report's recommendations:
- Ask before you assume. Replace "Did AI write this?" with "Walk me through how you made this."
- Make process part of the work. Before opening an AI tool, write four lines: the question, what I already know, what I think, what I still don't understand. On submission, add: which step used AI, and which AI suggestion I chose not to take.
The same holds for teams. When a leader openly shares where they used AI and how they checked it, others feel safe to do the same.
A searchlight can find a person. It cannot find a thought. Thoughts only become visible when people choose to share them.
Get the Human-First practice card
Four steps, ten minutes, with copy-ready AI prompts. The card and my weekly letter are currently written in Traditional Chinese.
👉 Subscribe free to 碩學苑 (Traditional Chinese)
Follow in English
I also post in English on LinkedIn.
👉 Follow Haishuo Lee on LinkedIn