“It is no longer the time to rely on individual teacher dedication. For written and essay assessments to be effectively implemented in education, a sophisticated system and public infrastructure at the national and educational authority levels must come first.”
On the afternoon of October 8, a forum titled ‘Exploring Expansion of Written and Essay Assessments’ was held at the National Assembly, where education policy officials and teachers gathered to discuss the urgent need for a paradigm shift in assessment to foster students' creative and critical thinking skills. They emphasized the necessity of establishing a concrete question and grading system, along with a Korean-style artificial intelligence (AI) support system.
Recently, the National Education Commission has been at the center of heated debates among educators and parents regarding the introduction of written and essay assessments in college entrance exams and the transition to absolute grading for both college entrance exams and school assessments. This forum, held at such a critical juncture, is seen as a pivotal moment for the future of college admissions and public education in South Korea, drawing significant public interest.
The event was co-hosted by the offices of National Assembly members Kang Deuk-gu and Kim Moon-soo, along with educational civic groups, and was supported by the Seoul, Gyeonggi, and Daejeon city education offices.
The forum, moderated by Kim Kyung-hee, head of the Curriculum and Assessment Innovation Research Division at the Korea Institute for Curriculum and Evaluation, moved beyond the traditional binary debate of ‘promoting private education’ or ‘grading unfairness’ to focus on practical alternatives that integrate the accumulated experiences from elementary, middle, and high school classrooms with AI and edtech.
In her opening remarks, Park Jong-im, a researcher at the Korea Institute for Curriculum and Evaluation, stressed that the fairness of written and essay assessments should not be solely focused on the consistency of scores among graders.
“In discussions about written and essay assessments, the focus is often on grading feasibility, but if the validity of the items collapses, fairness in grading cannot be established,” she stated. “The design of items must be closely aligned with national achievement standards, assessment objectives, command terms, and grading criteria.” She emphasized the need to move away from vague prompts like ‘explain’ or ‘write’ and clearly guide the cognitive level students are expected to demonstrate.
She also referenced examples from the UK A-Level, German Abitur, and the U.S. ETS, suggesting that for large-scale assessments, rather than relying solely on AI for automatic grading, a ‘human grading process’ utilizing benchmark responses and multiple grading should be designed first, with AI serving as a tool for anomaly detection.
Jung Sung-yoon, vice president of EduLift and a visiting professor at Jeonbuk National University, discussed the roadmap for introducing written and essay assessments in college entrance exams and the global standards for AI automatic essay scoring (AES).
“No large-scale assessments abroad use AI as a sole grader. In China’s Gaokao, subjective questions are entirely graded by humans, while AI is used for real-time monitoring of grader bias,” he noted, proposing a human-in-the-loop (HITL) mixed design as an alternative. He emphasized that in the introduction of a Korean-style college entrance exam, AI should handle initial grading and verification, while a certified grading panel composed of current teachers should make the final decisions, ensuring a minimum of 25% human review is legislated. He also stressed the need for institutional buffers to absorb shocks, such as transitioning to computer-based testing (CBT), moving to absolute grading, and addressing handwriting OCR errors.
During the designated discussion, teachers who have directly implemented written and essay assessments in elementary, middle, and high schools shared their firsthand experiences and policy challenges with education officials from city and provincial education offices.
Kim Ji-young, a supervisor at the Seoul Metropolitan Office of Education, emphasized that the HITL approach in schools should go beyond technical safety measures to ensure educational principles that guarantee teachers' professionalism. She stated, “To prevent teachers from uncritically following AI judgments, the Seoul Metropolitan Office of Education’s assessment support system, ‘Chaeum AI,’ is designed on the premise that teachers first read student responses and establish criteria for ‘pre-grading.’”
She introduced the ‘K-DOT (Depth of Thought-based Item Framework)’ developed by the Seoul Metropolitan Office of Education, advocating for a ‘shared responsibility system’ to prevent individual teachers from bearing the burden of complaints about assessment results, involving school subject councils and educational support evaluation committees.
Park Sang-wook, an English teacher at Daegu National University of Education’s affiliated high school, pointed out that the indirect writing questions in the college entrance exam do not accurately measure students' actual writing abilities. He shared his experiences over the past two years participating in the school-level backward design and ‘Daegu-style written and essay assessment’ as a senior grader.
“If sampling grading, consensus on detailed grading criteria, and continuous monitoring through benchmark responses are implemented, we can secure socially acceptable reliability,” he said, adding, “The issue is not the lack of teachers' capabilities but the absence of a system to support those capabilities. We should not only measure what is ‘easy to assess’ but also establish a national item pool and grading monitoring system that can accurately evaluate ‘educationally important competencies.’”
Lim Seong-eun, a senior teacher of English at Puruen Middle School in Hwaseong, noted that middle schools are already integrating written and essay assessments into daily classroom practices in line with the 2022 revised curriculum and achievement assessment system. She stated, “Simple memorization or rote learning cannot compete with the depth of thought required by written and essay assessments.”
Lim pointed out that a serious issue is the disconnect between the inquiry skills developed in middle school and the multiple-choice format of the college entrance exam, which leads to a ‘gap between school levels.’ She urged for the urgent integration of inquiry-based teaching and assessment, routines for feedback and revision, and the establishment of a public AI assessment support network to reduce regional disparities.
Jung Na-ra, a teacher at Hyochon Elementary School in Suwon, expressed concerns that while written and essay assessments are being attempted across all grades, the qualitative progress is lagging behind the quantitative expansion. She noted, “In math, a student may lose points for insufficient expression even if their reasoning is correct, or grading criteria may be vague, leading to discrepancies among graders. We need developmental stage-specific assessment guidelines that progress from early grades (words and pictures) to middle grades (explanations and reasons) to upper grades (logical expression and perspective acceptance), along with careful guidelines and public support to ensure that the introduction of AI tools does not double teachers' workloads.”
Following the designated discussion, a comprehensive discussion and audience Q&A session saw sharp questions and suggestions from teachers and education officials.
A representative from the National Education Commission raised concerns about the adjustment time required for new teachers with limited experience in written and essay assessments to collaborate with colleagues on question creation and grading, as well as the increased workload compared to multiple-choice assessments. In response, Park Sang-wook stated, “It took at least one semester of focused time to collaborate with fellow teachers to establish the assessment and grading process, and the initial workload was significant, providing detailed feedback of 510 minutes per student. However, in the formative assessment phase, using AI for linguistic feedback allowed us to focus on content feedback, significantly reducing time. Most importantly, we could endure it because we saw student growth.”
Lim Seong-eun also noted that practical training and intensive workshops offered in Gyeonggi Province have been very helpful, stating, “While the grading workload is substantial, I am experiencing educational benefits as students' blank responses are disappearing and class engagement is increasing.”
Direct criticisms of institutional contradictions in the education field were also voiced. Lee Soo-kyung, a supervisor at the Yongin Education Support Office, pointed out that even the process of receiving statements regarding school violence reveals that students lack basic literacy skills to articulate facts logically. She stated, “Parents and students are burdened with written and performance assessments throughout the year, and teachers are exhausted from excessive workloads. Simply adding written and essay assessments on top of existing multiple-choice assessments will only lead to backlash,” emphasizing that assessment reform should involve ‘changing’ rather than ‘adding’ to the existing evaluation system.
Kim Sung-dae, a teacher at Gangseo High School in Seoul, remarked, “Providing high-quality written and essay feedback to 28 students in a 50-minute class is nearly impossible.” He urged that relying solely on teachers' dedication to push policies and guidelines will only create confusion in the field, similar to the high school credit system. He requested that a warm and robust administrative and technical system be designed to allow teachers to confidently support student growth. In response, Kim Kyung-hee, head of the evaluation institute, acknowledged the importance of the point that support should come from a system rather than individual teacher dedication.
In her closing remarks, Jeong Mi-ra, director of the Educational Policy Design Institute and a member of the National Education Commission, highlighted that the fact that over 60% of students achieving the top grade in college entrance exams are repeat test-takers starkly illustrates the limitations of the existing multiple-choice assessment system. She stated, “Instead of getting caught up in the divisive debate over whether expanding written and essay assessments promotes private education, we must gather collective intelligence on what systems and technological enhancements we can implement to cultivate students' future thinking skills.”
Meanwhile, the lawmakers and education superintendents who co-hosted the forum pledged to provide policy support for the expansion of written and essay assessments.
Kang Deuk-gu, a member of the Democratic Party, stated in a written message, “To assess the ability to read materials, think critically, and express arguments, the questions and grading criteria must be clear, and students must be given ample opportunities to learn and practice in school. We will work together to design a system that supports schools and teachers to prevent disparities in educational opportunities.” He added that while AI should enhance the efficiency and consistency of assessments, trust must also be built regarding who will be responsible for the final results.
Kim Moon-soo, a member of the National Assembly Education Committee from the Democratic Party, also emphasized the need to reassess evaluations to help students actively interpret information and explain their judgments in an era where AI handles much of the thinking. He stated, “As various education offices in Seoul, Gyeonggi, Gangwon, and Daejeon are promoting the expansion of written and essay assessments, we will gather national solutions that encompass grading reliability, teacher workload, and system stability.”
Gyeonggi Province Education Superintendent Ahn Min-seok remarked in a written address, “Although it has been over 30 years since the college entrance exam began, students are still stuck in the practice of selecting a single correct answer, and simple memorization and calculation have already surpassed human capabilities with AI.” He added, “The education office will serve as a strong shield to ensure that teachers can make professional judgments and grade with confidence, emphasizing that while AI should be a tool to alleviate teachers' burdens, the essence of understanding students' thoughts must remain the responsibility of teachers. The evaluation in classrooms and the college entrance system must change together to complete the educational transformation.”
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.