Skip to main content Skip to navigation

Students' Perceptions and Preferences of Generative Artificial Intelligence Feedback for Programming

Project Overview

The document investigates the application of generative AI, particularly ChatGPT, in delivering automated feedback on programming assignments within an introductory computer science course. It highlights students' perceptions of this AI-generated feedback, assessing its adherence to formative feedback principles, their desire for personalized comments that reference their specific code, and recommendations for enhancement. The findings suggest that students regard the AI feedback as generally effective, appreciating its ability to provide timely assistance while expressing a strong preference for feedback that includes specific examples from their work to improve clarity and correctness. Overall, the use of generative AI in this educational context appears to enhance the learning experience by offering students constructive insights that can facilitate their understanding and development in programming.

Key Applications

Automated feedback generation using ChatGPT for Java programming assignments.

Context: Undergraduate introductory computer science course (CS1) at a U.S. public university.

Implementation: The ChatGPT API was used to generate feedback based on students' Java code for lab assignments, with surveys conducted to assess student perceptions and preferences.

Outcomes: Students perceived the feedback as aligning with formative feedback guidelines, preferring feedback that included their code for specificity and clarity.

Challenges: Challenges include ensuring the accuracy of AI feedback, managing the tone of feedback, and limitations in token counts affecting feedback depth.

Implementation Barriers

Technical Barrier

Token limits on ChatGPT can restrict the depth of feedback that can be generated for extensive code assignments.

Proposed Solutions: Future research could explore methods to effectively manage token usage or optimize feedback generation processes.

Ethical Barrier

Concerns about academic integrity and the potential for students to misuse AI tools for cheating.

Proposed Solutions: Developing guidelines and policies for responsible use of AI in education and incorporating AI as a teaching aid.

Project Team

Zhengdong Zhang

Researcher

Zihan Dong

Researcher

Yang Shi

Researcher

Noboru Matsuda

Researcher

Thomas Price

Researcher

Dongkuan Xu

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Zhengdong Zhang, Zihan Dong, Yang Shi, Noboru Matsuda, Thomas Price, Dongkuan Xu

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