In his People's Daily essay on AI and education, Shi Yigong argues students need to be freed from "mechanical memorisation through repetitive drilling" (重复刷题的机械记忆) and "the habitual mindset of chasing standard answers" (追求标准答案的惯性思维), using an interest-led approach to build the internal motivation to think independently and critically — specifically so they can construct their own path forward "in a world without standard answers" (在没有标准答案的世界里). That last phrase is doing a lot of work: it's naming the actual shift he thinks education needs to make, not just adding "critical thinking" to a list of nice-to-have skills.

The distinction that matters: retrieving an answer versus constructing one

Most KS3 assessment, reasonably, is built around problems that do have a standard answer — a correct method, a correct interpretation supported by the text, a correct historical explanation grounded in evidence. Learning to find that answer reliably is a real, necessary skill, and nothing in Shi Yigong's argument suggests otherwise.

The risk he's naming is what happens when every problem a student encounters is this kind — one with a predetermined correct answer that exists somewhere and just needs to be located or recalled. Practised exclusively, this builds a specific reflex: when faced with a question, look for the answer someone else already knows. That reflex works well right up until a student meets a genuine question that doesn't have a predetermined answer yet — which describes most of the interesting problems that exist outside formal education, and, per his broader argument, an increasing share of what an AI-saturated economy will actually need from people.

Why AI makes this specific gap more urgent, not less

It might seem like AI would make "finding the standard answer" more important, since AI is extremely good at retrieving known information. Shi Yigong's argument runs the opposite direction: because AI can now retrieve standard answers faster and more reliably than a person can, a person whose main trained skill is retrieving standard answers has little left to offer that AI doesn't already do better. What remains distinctively valuable is the capacity to work productively on questions that don't yet have an answer anywhere to retrieve — which is a different skill, and one that repetitive drilling toward known answers doesn't build.

Where "interest-led" comes in

Shi Yigong's proposed mechanism for building this is 兴趣先导 — an interest-first or interest-led approach, on the reasoning that genuine curiosity is what supplies the internal motivation needed to keep working on a hard, open-ended problem past the point where a quick lookup would resolve it. This connects directly to the argument in what AI can't do: curiosity and intrinsic drive aren't separate from critical thinking, they're the fuel that makes sustained critical thinking possible at all, especially on a question that resists quick resolution.

Why this is, structurally, the entire premise of Socratic tutoring

This is where Shi Yigong's argument connects most directly to how aitutors.me's tutors are built to behave. A tutor that responds to a stuck student by immediately supplying the correct method is training the exact reflex Shi Yigong warns against — reinforcing that the right move, when stuck, is to get the standard answer from an authoritative source as fast as possible. A tutor built on the Socratic method, deliberately never handing over the final answer and instead asking questions that lead a student to construct their own path to it, is a structural, repeated practice run at the opposite skill: building an answer, not retrieving one. See why tutors don't give answers for the mechanics of how this works in practice.

The connection isn't incidental. If Shi Yigong is right that the world increasingly rewards the capacity to think productively where no standard answer exists, then a tutoring method that trains a student to always expect the standard answer handed over on request is preparing them for exactly the wrong world — regardless of how efficiently it gets today's homework finished.

What this looks like for a parent, day to day

The practical version is resisting the urge — in yourself, not just in a tutoring tool — to supply the answer the moment a child is stuck, even when you know it and it would be faster. Asking "what have you tried" or "what would you guess, and why" before offering anything, even on ordinary homework questions that do have a standard answer, is a small, repeatable way to keep the construction muscle active rather than letting it atrophy in favour of the retrieval reflex.

FAQ

What does Shi Yigong say about 'standard answers'?

He argues students must be freed from mechanical memorisation through repetitive drilling and the habitual mindset of chasing standard answers, using interest-led learning to build the motivation to think independently and critically in a world without standard answers.

Isn't chasing the right answer the whole point of KS3 exams?

For exam technique, yes. Shi Yigong's concern is the mindset this can train if it's the only mode ever practised: a reflex to search for an answer someone else already knows, rather than the capacity to construct a defensible answer where none has been supplied.

How does this connect to how an AI tutor should actually respond to a student?

Directly — a tutor that immediately supplies the correct method reinforces the standard-answer reflex. A Socratic tutor that never hands over the final answer trains the opposite habit: constructing an answer rather than retrieving one.

Source

Adapted from 施一公 (Shi Yigong), "《人工智能时代，教育何为》" ("In the Age of AI, What Should Education Do?"), People's Daily, "大家手笔" column, 3 June 2026.

Duke Harewood · aitutors.me · Updated 14 Jul 2026.