Issues of Plagiarism and Moral Right to Attribution When Using Work Created by Large Language Models

Abstract

The advent of large language models such as ChatGPT has led people to use text responses generated by large language models (LLMs) in their research publications. This paper explores the issues arising from such use, including - originality of ideas and work, originality of text generated by LLMs, and plagiarism and moral right to attribution when work created by LLMs is used by authors. The paper proposes a plagiarism policy for using text generated by LLMs. Broadly, the policy proposes the following. If a work copies the response of LLMs, the use of LLM should be acknowledged, and the text should be presented in quotations. The original source of the information must be searched and acknowledged. If the work paraphrases the response of LLM, the use of LLM should be acknowledged. The original source of the information must be searched and acknowledged. If the work resulted from ideas generated during interaction with LLM, the use of LLM should be acknowledged for the help in generating the idea.