Shin, Jung, & Kwon (2026). 바이브 코딩 기반 수능영어 자동문항생성 앱 개발...

[Dongkwang Shin]

신동광, 정혜경, 권수현. (2026). 바이브 코딩 기반 수능영어 자동문항생성 앱 개발: One-shot 방식과 One-shot CoT 방식의 비교. 영어교과교육, 25(2), 127-146. 

Shin, Dongkwang, Jung, Hyekyung & Kwon, Su-hyeon (2026). Development of vibe coding-based automated CSAT English item generation applications: A comparison of One-shot and One-shot CoT approaches. Journal of the Korea English Education Society, 25(2), 127-146.

This study investigates the feasibility of developing automated CSAT English item generation (AIG) applications through “vibe coding”—a no-code, AI-assisted approach—and examines how prompting strategies influence item quality. Using Google AI Studio, two web-based AIG applications were built using One-shot and One-shot Chain-of-Thought (CoT) prompting. Sixty blank-inference items were automatically generated by the systems and eval‎uated by experts across five domains. The results confirm‎ that vibe coding enables non-programmers to design sophisticated assessment tools. While both methods produced pedagogically valid items, prompting strategies selectively influenced quality: One-shot prompting showed significantly better alignment with target difficulty, whereas One-shot CoT produced more plausible distractors. No significant difference was found in total scores or interaction effects. These findings suggest that vibe coding enhances teacher agency by democratizing AI tool development. The study concludes that prompting strategies should be strategically selected—One-shot for difficulty calibration and CoT for logical distractor construction—supporting more effective human-I collaborative assessment design.
[automated item generation/vibe coding/CSAT/prompting/자동문항생성/바이브 코딩/수능/프롬프팅]

06-신동광_정혜경_권수현-3교.pdf

739.82KB