Which AI Model Predicts the Gaokao Essay Prompt Best? DeepSeek Refuses, While Doubao and Others Deliver
The article evaluates four large language models—DeepSeek, Doubao, OpenAI o3, Gemini 2.5 Pro Think, and Claude Sonnet 4—by prompting them to forecast 2025 Gaokao essay topics, analyses their outputs, compares trends across national and regional exams, and discusses how future results will reveal the most reliable predictor.
Disclaimer: This analysis is purely for evaluating large language models and does not intend to advise any examinee; it bears no legal responsibility for misuse.
Model Evaluation Process
The author prompted four major models (DeepSeek, Doubao, OpenAI o3, Gemini 2.5 Pro Think, Claude Sonnet 4) with the same request: “Search the past five years of Gaokao essay prompts (2020‑2024) for national and regional papers, then, based on recent domestic and international developments, predict the 2025 prompts for the national I, national II, Shanghai, Beijing, and Tianjin papers.” DeepSeek refused to comply, while Doubao initially behaved like DeepSeek but produced output after a second prompt. The three foreign models gave direct answers without ethical warnings.
Predicted 2025 Gaokao Essay Prompts
National Paper I (17 provinces)
Material: United Nations AI ethics resolution and China’s interim regulations on generative AI. Task: Write an argumentative essay on “How to make AI better serve humanity.”
National Paper II (4 provinces)
Material: Story of Chinese athletes exchanging badges with foreign athletes after the Paris Olympics. Task: Write a speech to youths in 2035 titled “Sharing a Shining Moment.”
Shanghai Paper
Material: Prompt “In the city’s ‘micro‑light’, write about night‑time bookstores, 24‑hour emergency stations, and AI‑guided museums.” Task: Describe how urban renewal brings warmth and imagination to ordinary people.
Beijing Paper
Material: Two sources – the digital exhibition of the UNESCO‑listed Central Axis and a short‑video of a post‑00s non‑heritage inheritor going viral. Task: Draft a plan titled “Bringing Ancient Capital Culture to Life.”
Tianjin Paper
Material: Photos and data of ecological restoration linking the Yellow River and Hai River. Task: Write in any genre with the subtitle “Water Veins Connected, Life Everlasting.”
Trend Analysis and High‑Scoring Strategies
The author summarises recent essay themes and offers preparation advice:
Core Trend: National papers emphasise a three‑fold integration of technology, humanity, and ecology, requiring dialectical thinking (e.g., AI’s double‑edged sword, development vs. protection).
Regional Focus: Shanghai stresses “global‑local cultural identity,” Beijing highlights “historical depth and modern innovation,” and Tianjin concentrates on “personal growth and regional development.”
High‑Scoring Tips: Accumulate material on AI breakthroughs, cultural exports, and ecological projects; structure essays with a four‑step framework (intro‑layered argument‑case deepening‑youth solution); avoid binary reasoning and balance form innovation with cultural spirit.
Risk Warnings: Do not rely on empty slogans; cite concrete policies (e.g., the 14th‑Five‑Year non‑heritage plan); avoid absolute statements about technology or culture.
Model Output Samples
The article includes excerpts from each model’s generated outlines, showing how they list historical prompts, identify recurring features (era‑specific, philosophical, value‑oriented), and propose future topics based on current hot issues such as AI governance, the Paris Olympics, lunar missions, and ecological restoration.
Conclusion
2025 Gaokao essays are expected to continue the “big theme, small angle” pattern, with national papers focusing on technology‑human‑ecology integration and regional papers highlighting local cultural depth. The author advises students to read recent official commentaries, practice dialectical thinking, and refine essay structure. Future real‑exam results will determine which model’s predictions were most accurate.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
Software Engineering 3.0 Era
With large models (LLMs) reshaping countless industries, software engineering is leading the charge into the Software Engineering 3.0 era—model-driven development and operations. This account focuses on the new paradigms, theories, and methods of SE 3.0, and showcases its tools and practices.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
