Assessing the Reliability and Validity of Automated Feedback from ChatGPT in Programming Scenarios

Introduction: This study explores chatbot interactions in programming assessments to uncover distinct user personas and evaluate the reliability of ChatGPT-generated feedback.

Methodology:

• Data Extraction: Collect conversation metadata, linguistic features, sentiment cues, and engagement patterns.
• User Persona Analysis: Identify and classify behavioral profiles based on interaction styles.
• Performance Comparison: Evaluate student metrics before and after the integration of automated feedback.
• Synthesis: Combine findings to assess the educational impact and reliability of AI-based feedback.

Back to Theses Topics