Paper Number
2132
Paper Type
Complete Research Paper
Abstract
Generative AI has great potential for ubiquitous access to individualized instruction, which has important ramifications for equal access to education. However, educational theories suggest that constant guidance can impair deep thinking skills. We aimed to evaluate how to best provide AI support, hypothesizing that faded guidance facilitates better transfer learning. In a pre-registered experiment, students were randomly assigned to learn SQL with an AI tutor, providing either full access to assistance or access that faded. The results were compared on tasks assessing memorization and transfer skills. Contrary to the hypotheses, students supported by the non-restricted AI tutor demonstrated significantly stronger learning gains on memorization tasks. These findings offer novel insights into the role that generative AI plays in optimizing learning outcomes and how to implement it into the learning process.
Recommended Citation
Sengewald, Julian; Wilz, Mathis; and Lackes, Richard, "AI-Assisted Learning Feedback: Should Gen-AI Feedback Be Restricted to Improve Learning Success? A Pilot Study in a SQL Lecture" (2024). ECIS 2024 Proceedings. 12.
https://aisel.aisnet.org/ecis2024/track13_learning_teach/track13_learning_teach/12
AI-Assisted Learning Feedback: Should Gen-AI Feedback Be Restricted to Improve Learning Success? A Pilot Study in a SQL Lecture
Generative AI has great potential for ubiquitous access to individualized instruction, which has important ramifications for equal access to education. However, educational theories suggest that constant guidance can impair deep thinking skills. We aimed to evaluate how to best provide AI support, hypothesizing that faded guidance facilitates better transfer learning. In a pre-registered experiment, students were randomly assigned to learn SQL with an AI tutor, providing either full access to assistance or access that faded. The results were compared on tasks assessing memorization and transfer skills. Contrary to the hypotheses, students supported by the non-restricted AI tutor demonstrated significantly stronger learning gains on memorization tasks. These findings offer novel insights into the role that generative AI plays in optimizing learning outcomes and how to implement it into the learning process.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.