Skip to main content Skip to navigation

Can ChatGPT pass the Vietnamese National High School Graduation Examination?

Project Overview

The document examines the role of generative AI, particularly ChatGPT, in education, highlighting its performance on the Vietnamese National High School Graduation Examination (VNHSGE). It reports that ChatGPT successfully passed the exam, demonstrating proficiency in a range of subjects such as mathematics, English, physics, chemistry, biology, history, geography, and literature. These findings indicate the promising potential of AI tools to enhance educational experiences. However, the research also underscores the necessity for further exploration of ChatGPT's effectiveness with more complex questions and its applicability across diverse educational settings. Overall, the document emphasizes the transformative potential of AI in education, suggesting that generative AI can significantly benefit both students and educators by improving learning outcomes and engagement.

Key Applications

ChatGPT used to complete the Vietnamese National High School Graduation Examination (VNHSGE)

Context: High school education in Vietnam; target audience includes students preparing for graduation exams.

Implementation: ChatGPT was evaluated on a dataset specifically created from VNHSGE exam questions, including essays and multiple-choice questions.

Outcomes: ChatGPT passed the VNHSGE with an average score of 6-7, demonstrating proficiency in various subjects.

Challenges: Performance varies by subject; some subjects revealed lower scores, indicating a need for improvement in understanding complex questions.

Implementation Barriers

Technical Limitations

ChatGPT struggles with high-order thinking and analytical skills required for solving complex exam questions.

Proposed Solutions: Further research into enhancing AI capabilities for critical thinking and problem-solving tasks is needed.

Contextual Understanding

ChatGPT's performance can differ significantly based on the complexity of questions and cultural context.

Proposed Solutions: Expand training data to include more contextually rich examples and complex exam questions.

Project Team

Xuan-Quy Dao

Researcher

Ngoc-Bich Le

Researcher

Xuan-Dung Phan

Researcher

Bac-Bien Ngo

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Xuan-Quy Dao, Ngoc-Bich Le, Xuan-Dung Phan, Bac-Bien Ngo

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies