SlideItRight: Using AI to Find Relevant Slides and Provide Feedback for Open-Ended Questions
Project Overview
The document discusses the implementation of SlideItRight, a cutting-edge system that leverages generative AI to provide personalized textual feedback alongside relevant lecture slides, aiming to enhance student learning experiences. By employing Large Language Models (LLMs), the system delivers multimodal feedback that is contextually tailored to students' needs. The findings demonstrate that while there are notable learning gains across various feedback types, no significant differences were observed between them. Student feedback indicates a positive reception towards the personalization offered by AI-generated comments; however, concerns regarding trust and clarity compared to traditional human feedback were noted. Ultimately, the study concludes that while AI can effectively bolster educational feedback mechanisms, considerations regarding cognitive load and the clarity of the information provided are critical for its successful integration into educational settings.
Key Applications
SlideItRight - AI feedback system that integrates textual feedback with relevant slides.
Context: Online learning environment targeting university students.
Implementation: Implemented through a crowdsourcing study with 91 participants, comparing AI-generated feedback with human feedback and relevant slide retrieval.
Outcomes: Significant pre-to-post learning gains observed across all conditions; students found AI feedback personalized but had lower trust compared to human feedback.
Challenges: Students reported difficulty understanding AI feedback and expressed lower trust in its reliability.
Implementation Barriers
Trust and Reliability
Students expressed concerns about the accuracy and reliability of AI-generated feedback.
Proposed Solutions: Enhance AI feedback with explicit justifications, clear rubrics, and inline references linking to reliable course materials.
Cognitive Load
The combined feedback from AI and slides sometimes overwhelmed students, leading to cognitive overload.
Proposed Solutions: Implement strategies to manage information density, such as simplifying language, providing multi-level hints, and ensuring clear presentation of information.
Project Team
Chloe Qianhui Zhao
Researcher
Jie Cao
Researcher
Eason Chen
Researcher
Kenneth R. Koedinger
Researcher
Jionghao Lin
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Chloe Qianhui Zhao, Jie Cao, Eason Chen, Kenneth R. Koedinger, Jionghao Lin
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