Skip to main content Skip to navigation

Exploring the Role of AI Assistants in Computer Science Education: Methods, Implications, and Instructor Perspectives

Project Overview

The document examines the integration of generative AI, specifically ChatGPT, in computer science education, emphasizing its potential to enrich learning experiences while also presenting challenges such as academic integrity issues, compromised problem-solving skills, and the formation of misleading mental models among students. It assesses ChatGPT's effectiveness in addressing various computer science problems and discusses strategies to adapt these problems to mitigate misuse. Through interviews with computer science instructors, the study uncovers varied perspectives on AI's influence, highlighting a shared recognition of the necessity for adaptive teaching methods and innovative tools to support the effective incorporation of AI into educational practices. The findings suggest that while generative AI can enhance learning, careful consideration and strategic implementation are essential to address the associated risks and maximize its benefits in educational contexts.

Key Applications

ChatGPT as an AI assistant for problem-solving in computer science courses

Context: Undergraduate computer science courses across various topics and levels

Implementation: Evaluated ChatGPT's performance on problems from selected courses and explored modification methods to adapt problems for AI use

Outcomes: ChatGPT solved 60.85% of the problems; insights on potential misuse and instructor adaptation strategies

Challenges: Concerns about academic integrity, students' reliance on AI for problem-solving, and the generation of incorrect answers leading to flawed mental models

Implementation Barriers

Ethical

Concerns regarding academic integrity and potential for cheating due to AI's problem-solving capabilities

Proposed Solutions: Develop strategies for instructors to modify course materials and assessment methods to mitigate misuse

Implementation

Instructors find it challenging to adapt courses and materials to incorporate AI tools effectively

Proposed Solutions: Provide support tools and guidance for instructors to modify materials and implement AI in their courses

Technical

ChatGPT's outputs can be inaccurate, leading to the development of incorrect mental models among students. It is important to educate students on verifying information provided by AI and creating awareness of its limitations.

Proposed Solutions: Educate students on verifying information provided by AI and creating awareness of its limitations.

Project Team

Tianjia Wang

Researcher

Daniel Vargas-Díaz

Researcher

Chris Brown

Researcher

Yan Chen

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Tianjia Wang, Daniel Vargas-Díaz, Chris Brown, Yan Chen

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