ChatGPT-Assisted Retrieval Practice and Exam Scores: Does It Work?
This paper examines the potential of ChatGPT-assisted retrieval practice to enhance students’ final exam performance. ChatGPT was utilized to generate questions and deliver timely feedback during retrieval practice, supporting learning in large class settings where providing personalized feedback is often challenging.
Students often excel in continuous assessments yet face significant challenges in final exams, largely due to the demanding nature of these exams that require the recall and application of accumulated knowledge. This persistent issue highlights a gap in traditional study practices and underscores the need for innovative strategies to support long-term memory retention. This study explores how ChatGPT can bridge this gap by supporting retrieval practice, an evidence-based strategy known to improve long-term memory retention.
This study adopts a retrospective cohort design, comparing final exam scores between previous cohorts who did not use ChatGPT for retrieval practice (control group) and current cohorts who did (experimental group). ChatGPT was used to generate objective questions for retrieval practice and provide immediate feedback to students. The primary sample consists of second-year education students enrolled in the Measurement and Evaluation in Education course, with 64 students randomly selected for each group. In addition to analyzing exam scores, the study incorporates complementary findings from students’ feedback collected at the end of the semester to gain a deeper understanding of their experiences with ChatGPT-assisted retrieval practice.
As higher education in Malaysia increasingly shifts towards alternative assessments, this study highlights a simple yet impactful retrieval practice as a learning strategy that integrates formative assessment with feedback. With the aid of generative AI such as ChatGPT, such strategies can be implemented easily and effectively by offering a practical solution that enhances exam performance while reducing the lecturer’s workload.
The results of the study indicate that students who engaged in AI-assisted retrieval practice using ChatGPT performed significantly better on their final exams compared to those who did not. A Welch’s t-test revealed a significant difference in exam scores between the control group (students without ChatGPT-assisted practice) and the experimental group (students with ChatGPT-assisted practice). While the findings demonstrate the effectiveness of ChatGPT in enhancing academic performance, they also underscore the importance of complementing AI support with human feedback to address complex learning needs and provide deeper contextual understanding.
Practitioners should consider using AI tools like ChatGPT to support retrieval practice by helping students generate questions and receive immediate feedback, particularly in large classes. AI feedback should be complemented with human input to address complex questions and provide deeper understanding. Providing students with guidance on using AI tools effectively can maximize their benefits while balancing group activities with individual tasks ensures both collaboration and personalized engagement. Regularly gathering student feedback and tracking performance can help refine the integration of AI tools to better meet learning goals.
Future studies should investigate the long-term impact of ChatGPT-assisted retrieval practice on memory retention and explore how different types of ChatGPT-generated feedback influence learning outcomes. Additionally, the research could examine how the integration of lecturer-generated or student-generated questions in ChatGPT-assisted retrieval practice affects learning and whether varying these approaches could further enhance memory retention and exam performance across different educational contexts and subjects.
ChatGPT-assisted retrieval practice, when effectively implemented, transforms retrieval practice from being just an assessment tool into a powerful learning strategy that enhances memory retention. With ChatGPT providing timely feedback and supporting educators and students, it can reduce the burden on educators in large classrooms while empowering students to take greater control of their learning.
Future research should explore the longer and more frequent use of ChatGPT-assisted retrieval practice to give students more opportunities for individual engagement. Studies could also focus on improving ChatGPT’s feedback for more complex questions and finding the best ways to combine AI feedback with human input. Expanding research into different subjects and tracking long-term effects on learning and performance would help better understand how ChatGPT can be effectively used in education.
Back