On June 7, 2026, the National College Entrance Examination officially kicked off, with 12.9 million candidates heading to the examination halls. The most anticipated essay topic for the Chinese language section was revealed as expected. The micro-writing question in the Beijing volume focused on the theme of "Artificial Intelligence and a Happy Senior Life," while the Shanghai volume posed a philosophical question about "the transformation of the world by technology and the change in human imagination."

Regarding these two exam questions that closely follow technological trends, according to the latest industry large model ranking, three top AI models in China—DeepSeek, Tongyi Qianwen, and Douyin—competed together and submitted their own answers. In the micro-writing section of the Beijing volume, all three models used delicate and warm language to create vivid promotional slogans for elderly activities. When facing the in-depth reasoning of the Shanghai volume, the large models each showcased their strengths. DeepSeek believed that technology is a catalyst for the rebirth of imagination, Qianwen warned against human thinking being regulated by algorithms, and Douyin advocated expanding cognitive boundaries through technology and illuminating the future with imagination.

image.png

However, the writing skills demonstrated by the large models in the "exam hall" could not hide their complete failure in "prediction" before the exam. Just before the Gaokao, some self-media accounts used ChatGPT, Claude, Gemini, Douyin, and Kimi, five mainstream models, to predict the composition topics of the 2026 Gaokao under the same prompt. Looking at the directions given by each model, Gemini bet on "a frictionless life" for technological reflection, Kimi focused on "path confusion under algorithm navigation," while Douyin, ChatGPT, and Claude also proposed topics from perspectives such as "the era of shortcuts," "measures," and "scenery on the way," covering technological ethics and grand narratives.

Today, when the real exam topics were released, none of the predictions made by these five models hit the mark. For example, the essay topic for the national volume one did not follow the technological ethics debate predicted by AI, but instead provided a question filled with human warmth and growth memories, asking candidates to discuss their understanding of a certain word that has changed, based on their personal growth.

In response to the recent phenomenon of using AI to hype up Gaokao prediction topics in the market, the Ministry of Education has recently issued the "2026 Gaokao Warning Information." The official clearly stated that as the Gaokao approaches, some unscrupulous individuals and professional institutions are taking advantage of the desire of students and parents to gain an unfair advantage, promoting false advertising under the guise of famous teachers or AI prediction topics, and even using pre-sales, time-limited offers, and malicious links to lure people to purchase at high prices, committing fraud.

The Ministry of Education specifically reminds that in recent years, the Gaokao has been continuously deepening its reform, with innovative directions and content, focusing on examining key abilities and thinking qualities of candidates, which inherently has a clear "anti-prediction" and "anti-formula" orientation. Relying on AI or so-called expert predictions to achieve high scores is completely unrealistic. Candidates and parents should remain rational, avoid disrupting normal study routines due to blind pursuit of "prediction papers," and prevent both economic and psychological losses.