AI-Assisted OCR Grading Pilot for Vietnam's Restructured 2025 National High School Graduation Exam

Vietnam's 2025 restructuring of its National High School Graduation Exam introduced three question formats (multiple-choice, true/false and short-answer), moving away from pure optical-mark-recognition (OMR) bubble sheets that are prone to mechanical misreads. Researchers at Thái Nguyên University of Education (Tran Quang Hieu, Nguyen Thi Phuong and Tran Thi Thu) developed and piloted an AI-assisted grading pipeline that combines a direct-answer sheet format with Mathpix optical character recognition (OCR) to read students' handwritten short answers.
Using a five-stage Borg & Gall development model — regulatory analysis, system development, expert validation, dual-phase trials and pilot testing — the team tested the tool on 93 real student answer papers, achieving 98.5% character-recognition accuracy and user-satisfaction ratings of 4.0+ out of 5 among evaluators. The study was published in the peer-reviewed European Journal of Education Studies (Vol 12, No 4, 2025).
The pilot is small and has not been adopted for live grading of the exam, which in 2026 was sat by more than 1.2 million students at over 2,500 test sites — the largest such exam Vietnam has run. Separately, national authorities have flagged AI-enabled cheating (alongside hidden cameras, earpieces and impersonation) as a growing security concern and have responded with police coordination, staff and student awareness campaigns, and technical-detection pilots at high-risk sites, rather than with an AI-based grading or integrity system of their own. The OCR-grading research therefore represents an early, unscaled academic proof-of-concept rather than a nationally deployed system, and its 93-paper sample should not be read as validated at exam scale.

Read the full analysis: http://dx.doi.org/10.46827/ejes.v12i4.6007