PlanGlow: Personalized Study Planning with an Explainable and Controllable LLM-Driven System
Project Overview
The document highlights the use of generative AI in education, focusing on PlanGlow, an LLM-based system that creates personalized study plans for self-directed learners. It underscores the critical role of explainability and controllability in AI tools, which are essential for building trust and enhancing user experience. By allowing users to input their learning preferences, PlanGlow generates structured study plans accompanied by clear justifications for the chosen content, thereby fostering a deeper understanding of the learning process. Evaluations reveal that PlanGlow significantly outperforms traditional systems in terms of usability, explainability, and controllability, suggesting its effectiveness in enhancing personalized learning experiences. Overall, the findings indicate that generative AI has the potential to transform educational practices by providing tailored support to learners and improving their engagement and outcomes.
Key Applications
PlanGlow: Personalized Study Planning with an Explainable and Controllable LLM-Driven System
Context: The application is intended for self-directed learners seeking to create personalized study plans.
Implementation: Users input their learning preferences, and the system generates structured study plans using an LLM with integrated explainability and controllability features.
Outcomes: PlanGlow enhances usability, explainability, and controllability, leading to more informed decision-making and improved user engagement.
Challenges: Challenges include ensuring transparency in AI recommendations and mitigating the impact of hallucinated information.
Implementation Barriers
Technical Barrier
Lack of transparency in AI recommendations and potential for hallucinated information can confuse learners.
Proposed Solutions: Implementing explainability features to clarify AI reasoning and allowing users to adjust outputs.
Resource Limitation
Dependence on YouTube as the primary resource limits diversity and may not cater to all learning preferences.
Proposed Solutions: Incorporate a wider range of educational resources including reading materials and exercises.
Project Team
Jiwon Chun
Researcher
Yankun Zhao
Researcher
Hanlin Chen
Researcher
Meng Xia
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Jiwon Chun, Yankun Zhao, Hanlin Chen, Meng Xia
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