Skip to main content Skip to navigation

Generating Situated Reflection Triggers about Alternative Solution Paths: A Case Study of Generative AI for Computer-Supported Collaborative Learning

Project Overview

The document explores the integration of generative AI, particularly Large Language Models (LLMs) such as ChatGPT, within computer-supported collaborative learning (CSCL) settings, focusing on their application in a Cloud Computing course. It emphasizes the generation of contextualized reflection triggers aimed at improving student engagement and learning during collaborative programming tasks. The study assesses the impact of these AI-generated prompts on students' task completion rates and overall learning outcomes. Although the findings indicate that the reflection triggers did positively affect task completion, they did not lead to a significant improvement in learning outcomes when compared to a control group. This suggests that while generative AI can enhance specific aspects of student engagement in collaborative learning environments, its efficacy in boosting comprehensive learning results may require further exploration and refinement.

Key Applications

Integration of ChatGPT to provide dynamic and contextualized feedback during collaborative programming tasks.

Context: College-level Cloud Computing course for students working in groups.

Implementation: Augmented an Online Programming Exercise bot with ChatGPT to generate reflection triggers based on students' interactions and discussions.

Outcomes: Improved engagement through contextualized feedback; increased task completion rates for difficult tasks.

Challenges: Initial confusion from students regarding reflection triggers; no significant improvement in overall learning outcomes.

Implementation Barriers

Engagement Barrier

Students found some reflection triggers confusing and felt they did not always relate to their discussions.

Proposed Solutions: Future iterations will include more cohesive prompts that relate directly to student discussions and break down prompts into smaller, more readable parts.

Implementation Barrier

The reflection triggers generated sometimes provided suboptimal alternatives that did not enhance discussion.

Proposed Solutions: Improving the generation of alternative scenarios that are more relevant and optimal for the context of the tasks.

Project Team

Atharva Naik

Researcher

Jessica Ruhan Yin

Researcher

Anusha Kamath

Researcher

Qianou Ma

Researcher

Sherry Tongshuang Wu

Researcher

Charles Murray

Researcher

Christopher Bogart

Researcher

Majd Sakr

Researcher

Carolyn P. Rose

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Atharva Naik, Jessica Ruhan Yin, Anusha Kamath, Qianou Ma, Sherry Tongshuang Wu, Charles Murray, Christopher Bogart, Majd Sakr, Carolyn P. Rose

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

Let us know you agree to cookies