Is ChatGPT 3 safe for students?
Project Overview
The document explores the implications of using generative AI, specifically ChatGPT-3, in educational settings, particularly for coding assignments. While ChatGPT-3 demonstrates proficiency in generating correct code for standard programming tasks, the findings indicate significant risks associated with its use, particularly concerning academic integrity. A study revealed that 38% of the code produced by ChatGPT-3 matched existing code identified by a plagiarism detection tool, and a staggering 96% of the time, similar code could be located via Google searches. These results underscore the potential for high rates of plagiarism, prompting a critical need for educators to revise their teaching strategies and assessment methods to mitigate these risks. Overall, while generative AI tools like ChatGPT-3 have the capacity to enhance learning, their implementation in education must be approached with caution to preserve the integrity of academic work.
Key Applications
ChatGPT-3
Context: Programming assignments for students, specifically in the context of coding algorithms.
Implementation: Students were tasked with coding standard algorithms in Python using prompts directed at ChatGPT-3. The generated codes were then assessed for plagiarism using Codequiry and Google searches.
Outcomes: ChatGPT-3 produced correct code 100% of the time for basic requests, but high rates of similarity to existing code raised concerns about safety in terms of plagiarism.
Challenges: High likelihood of plagiarism detection when using ChatGPT-3 for coding assignments, with significant similarities found between generated and existing code.
Implementation Barriers
Academic Integrity
The use of generative AI like ChatGPT-3 can lead to plagiarism, as the generated code can closely resemble existing code.
Proposed Solutions: Educators need to adapt teaching methods and assessment practices to address the challenges posed by AI-generated content.
Project Team
Julia Kotovich
Researcher
Manuel Oriol
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Julia Kotovich, Manuel Oriol
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