Human–AI collaboration brings detailed feedback to every olympiad student

The problem: grading olympiad work has always been the heaviest job

Math olympiad problems are hard. A single proof can follow many approaches, with long chains of reasoning and plenty of detail. Reading every step of every student’s work and pointing out exactly what went wrong takes coaches an enormous amount of time.

With many students and difficult problems, detailed feedback rarely reaches everyone. It has long been one of the biggest pain points in olympiad teaching.

Step one: helping coaches use AI well

We ran AI training for the Guangdong Olympiad School’s mathematics coaches, covering lesson preparation, teaching and grading, so that AI became part of their everyday work. Afterwards, the coaches could use AI with much more confidence to prepare, teach and grade.

Step two: detailed feedback for every student

In OneAI, the AI drafts a score for each rubric point and explains why each mark was earned or lost. Coaches then confirm or adjust each question and set the final score.

Detailed feedback that once only a few students could receive now reaches every student on the training team. Students can see exactly which step went wrong, and coaches can spend more of their time explaining and guiding.

The strength of human–AI collaboration

AI handles a fast, detailed first pass; coaches bring professional judgment and teach each student well. Each plays to its strengths, which is exactly what we believe in: AI empowers teaching and learning, and the final decision always rests with the teacher.