In his People's Daily essay on AI and education, Shi Yigong names four qualities he says are distinctively human and that education must foreground as AI takes over standardised, knowledge-based work: 好奇心 (curiosity), 同理心 (empathy), 创造力 (creativity), and 内驱力 (intrinsic drive). His framing is specific: education should "more fully respect human subjectivity and manifest human value" — a direct argument that these four aren't soft extras alongside the "real" academic content, but the actual point of what education is for once knowledge-holding stops being a distinctively human advantage.

Why these four, and not a longer list

It would be easy to read this as a version of the familiar "21st century skills" list — communication, collaboration, critical thinking, and so on — but Shi Yigong's four have a narrower, sharper logic behind them. Each names something generative: curiosity is wanting to know something that isn't yet known to you; empathy is genuinely engaging with another person's experience; creativity is making something that didn't exist before; intrinsic drive is wanting to do something for reasons that come from inside you, not from an external requirement. None of these are things a system can have simply by having access to a large amount of existing information — they're not retrieval problems. That's precisely why they resist replication by a technology whose core strength, per the industrial-era argument, is retrieving and recombining what already exists.

Curiosity: the one that gets squeezed out first

Of the four, curiosity is the most fragile under exam pressure, and the easiest for a busy household to accidentally deprioritise. A KS3 schedule built entirely around covering required content on a timeline has no natural room for a question that isn't on the syllabus — and a student who learns, repeatedly, that off-syllabus questions get redirected rather than explored, reasonably learns to stop asking them. Protecting some space for a genuinely curious question to actually be followed, even briefly, is a direct, practical way to act on this part of Shi Yigong's argument.

Empathy: not a "soft skill" bolted onto the real subjects

It's tempting to file empathy under pastoral care, separate from academic learning. Shi Yigong's framing resists that separation — empathy, in his argument, is one of the specific things that makes human understanding different from AI's. In practice, this shows up in subjects that get treated as more "content-heavy" than they need to be: history taught as dates and outcomes rather than as an exercise in genuinely understanding why people in the past acted as they did; literature taught as identifying techniques rather than engaging with what a character or a writer actually felt and why. The content coverage doesn't have to change for this shift to happen — the emphasis within it does.

Creativity: harder to grade, which is exactly the problem

Creativity is disproportionately squeezed out of exam-focused subjects precisely because it's harder to grade reliably than a right-or-wrong answer. This creates a structural bias: assessment systems reward what's easy to mark consistently, which pushes both teaching and studying toward exactly the kind of standardised, convergent thinking Shi Yigong argues AI already does better. Making room for genuinely creative attempts — even in subjects like maths, where an unconventional method attempted and explained is worth more than the same right answer arrived at by rote — is a direct counterweight.

Intrinsic drive: the one that determines whether the other three survive

Of the four, intrinsic drive is arguably the load-bearing one, because curiosity, empathy, and creativity all require a student to actually want to engage rather than being pushed through the motions. This connects directly to self-efficacy — real confidence built from genuine accomplishment, not praise — because a student who's only ever externally driven (do this because it's due, because it's graded, because a parent said so) has no internal reserve to draw on once the external pressure eases, which is exactly the kind of motivation an AI-saturated world, per Shi Yigong's argument, will reward least.

What this looks like as an actual household practice, not just a value statement

None of these four qualities are taught through a single conversation or unit — they're built through accumulated small choices about what gets rewarded and what gets redirected. Curiosity gets built by occasionally following a tangent instead of redirecting to the syllabus. Empathy gets built by asking "why do you think they felt that way" rather than only "what happened next." Creativity gets built by valuing an interesting wrong attempt over a boring correct one, at least some of the time. Intrinsic drive gets built by letting a student's own reasons for caring about something matter, rather than every task being framed purely in terms of external requirement.

FAQ

What four qualities does Shi Yigong say education should emphasise?

Curiosity (好奇心), empathy (同理心), creativity (创造力), and intrinsic drive (内驱力) — arguing education must "more fully respect human subjectivity and manifest human value" as AI takes over standardised, knowledge-recall tasks.

Why these four specifically, rather than a longer list of '21st century skills'?

Each names something generative — wanting to know, caring about another person, making something new, wanting to for one's own reasons — rather than something retrievable, which is precisely what resists replication by a system built on retrieving existing knowledge.

Can these qualities actually be taught, or are they just personality traits?

Shi Yigong's framing treats them as things education can foreground or neglect, not fixed traits. A curriculum and household culture that consistently rewards curiosity, empathy, original attempts, and self-directed effort develops more of all four than one that never asks for them.

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.