Generative AI Usage and Exam Performance
Project Overview
The document explores the role of generative AI, particularly ChatGPT, in higher education, revealing its dual nature as both a learning facilitator and a potential hindrance to academic performance. While these tools can enhance engagement and provide valuable support in learning, the findings indicate a concerning correlation between their usage and lower exam scores, especially among more capable students. This suggests that reliance on generative AI may inadvertently undermine critical thinking and learning processes. The paper urges educators and academic institutions to critically assess the implications of integrating generative AI into educational environments, advocating for a balanced approach that harnesses its benefits while mitigating adverse effects on student outcomes.
Key Applications
ChatGPT and other generative AI tools
Context: Higher education, particularly in an introductory financial accounting course
Implementation: Students used generative AI for writing case study essays, with their performance analyzed via regression analysis comparing GenAI users and non-users.
Outcomes: GenAI users scored on average 6.71 points lower on exams than non-users. The findings suggest a learning-hindering effect, especially for students with high learning potential.
Challenges: The reliance on generative AI might hinder students' ability to think critically and independently, leading to superficial learning and lower exam performance.
Implementation Barriers
Educational Policy
Some educational institutions have prohibited the use of generative AI tools, fearing they undermine academic integrity and learning processes.
Proposed Solutions: Educational institutions should guide the proper use of generative AI tools, rather than banning them outright, to strike a balance between leveraging their benefits and mitigating negative effects.
Project Team
Janik Ole Wecks
Researcher
Johannes Voshaar
Researcher
Benedikt Jost Plate
Researcher
Jochen Zimmermann
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Janik Ole Wecks, Johannes Voshaar, Benedikt Jost Plate, Jochen Zimmermann
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