It's time our exams caught up with the future

When it comes to exams, let's stop pretending that AI doesn't exist. Instead, test students on how well they can use it.

- Looi Chee Kit and Wong Lung Hsiang

Every June in China, nearly 13 million students sit the gaokao, a high-stakes college entrance exam often dubbed the "most competitive exam in the world".

In 2025, however, something unusual happened: Generative artificial intelligence (AI) tools vanished during the exam, and it was on purpose.

Major Chinese tech companies like ByteDance, Alibaba and Tencent all voluntarily disabled or restricted their AI services during the exam period. AI chatbot tools refused to answer questions related to the exam syllabus, image recognition functions were suspended and, in some cases, entire services went offline.

The move was widely interpreted as a coordinated act—it was less about plugging genuine loopholes in the system, and more a performative signal: that AI should stay out of this sacred domain of meritocratic assessment.

Students were told, symbolically: "You're being watched." Platforms were reminded: "You are not above regulation."

Why such a strong response?

Because examinations in China, and across much of Asia, are more than just administrative procedures or academic formalities. They are deeply symbolic rituals, bound up with ideas of fairness, meritocracy and upward mobility.

For generations, these high-stakes tests have been seen as the great equaliser, where academic strength is earned through effort, not inherited privilege.

The mere presence of AI risks unsettling those assumptions, raising uncomfortable questions about what counts as genuine ability, and who gets to define it.

As global headlines from The Washington Post to The Guardian have noted, this also opens up a deeper, more uncomfortable question: Are we protecting examinations from AI, or shielding examinations from change?