Scholarlink

Multiple-choice questions (MCQs) are widely used in educational settings due to their automated rating capabilities, ease of administration, and reliability in evaluation. A crucial element of multiple-choice questions (MCQs) is distractors—incorrect answer choices that are designed to assess students’ conceptual understanding by seeming plausible while remaining incorrect. Advanced AI models, including BERT, ChatGPT, Llama, and Gemini AI, offer innovative approaches to generating and evaluating distractors. This study compares the effectiveness of distractor scores generated by advanced AI models, including BERT, ChatGPT, Llama, and Gemini AI, in multiple-choice questions. These AI-driven methods offer diverse approaches to improving the quality and sophistication of distractors, contributing to more robust and dynamic educational assessments. Distractors’ scores can provide educators with a valuable framework to quickly validate solutions, offering insight by analyzing the quantitative differences between distractors.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Distractor Analytics on Multiple-Choice Questions using Large Language Models

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Basavaraj N. Hiremath,
  • Savitha Hiremath

摘要

Multiple-choice questions (MCQs) are widely used in educational settings due to their automated rating capabilities, ease of administration, and reliability in evaluation. A crucial element of multiple-choice questions (MCQs) is distractors—incorrect answer choices that are designed to assess students’ conceptual understanding by seeming plausible while remaining incorrect. Advanced AI models, including BERT, ChatGPT, Llama, and Gemini AI, offer innovative approaches to generating and evaluating distractors. This study compares the effectiveness of distractor scores generated by advanced AI models, including BERT, ChatGPT, Llama, and Gemini AI, in multiple-choice questions. These AI-driven methods offer diverse approaches to improving the quality and sophistication of distractors, contributing to more robust and dynamic educational assessments. Distractors’ scores can provide educators with a valuable framework to quickly validate solutions, offering insight by analyzing the quantitative differences between distractors.