First Published on 16 July 2023.

This research paper presents a comprehensive study on the ethical principles and considerations when using ChatGPT in software engineering (SE) research. The study consists of three main parts: identification of motivators and demotivators, modeling of ethical principles using the Interpretive Structural Modeling (ISM) approach, and analysis using the Matrix Cross-Impact Multiplication Applied to Classification (MICMAC) method.

In the first part, motivators and demotivators of using ChatGPT in SE research are identified through a survey of 113 researchers. The study reveals various motivators such as code summarization and requirements description generation, which enhance efficiency and collaboration in the research process. Additionally, demotivators such as convention misalignment and model limitations are identified as potential challenges that could hinder ethical practices.

The second part of the study employs the ISM approach to model the identified ethical principles into five distinct levels. These levels provide a deeper understanding of the relationships among the principles, enabling more responsible and ethical use of ChatGPT in SE research. The analysis reveals the importance of addressing issues such as bias, privacy, reliability, and transparency to ensure ethical research practices.

The final part of the study utilizes the MICMAC analysis to categorize the ethical principles into four clusters based on their driving and dependence power. This analysis helps researchers identify key principles that have a significant impact on ethical considerations and guide decision-making in using ChatGPT. It highlights the importance of principles such as accountability, trust, fairness, and informed consent in ensuring responsible and ethical SE research.

Find the paper linked here: https://arxiv.org/pdf/2306.07557.pdf