Experiences of Academics, Graduates, and Undergraduates in Using Generative AI in Research (Un)Ethically and (Ir)Responsibly: A Systematic Review of Qualitative Synthesis

Synthesizing qualitative evidence on the experiences and perceptions of students and academics regarding generative AI usage in academic research, this review illuminates the ethical dimensions amidst growing application of AI tools like ChatGPT, Claude, and Gemini. By exploring data privacy, bias, and scholarly integrity concerns, it informs policy development and fosters responsible research practices involving generative AI. To evaluate the ethical considerations and integrity experiences of undergraduate and graduate students, academics, and faculty when using generative AI in research activities within higher education institutions worldwide. Searches were conducted in April 2024 across databases including Web of Science, Scopus, and PubMed with no language restrictions. Study selection involved title/abstract screening then full-text review. Critical appraisal assessed quality and relevance. Data extraction captured study details like populations and methods. A meta-aggregation approach synthesized findings into statements. Confidence was assessed for methodological robustness. Qualitative studies involving undergraduate/graduate students and academics/faculty at higher education institutions globally were included. Phenomena of interest were experiences and perceptions of ethics and integrity when using AI in research. Studies had to be conducted in academic research contexts using qualitative methods like phenomenology, grounded theory, ethnography, and action research. From 562 records, 30 studies were included representing diverse countries. Methodological quality was mixed; 55.9% rated high quality. Sixty findings were categorized into 10 areas then synthesized into 4 meta-aggregative flowcharts: 1) Need for ethical AI guidelines and policies; 2) AI impact on student learning and critical thinking; 3) AI detection and accuracy limitations; 4) Broader ethical and societal implications. Ambiguities exist around defining AI-related academic misconduct. While enhancing learning efficiency, AI raises concerns about diminished critical thinking from over-reliance. Robust AI detection and accuracy methods are needed. A holistic approach considering societal impacts is crucial for ethical AI integration in academia. Recommendations include developing clear policies, educational interventions, and rigorous longitudinal research. Systematic Review Registration: This review protocol was registered on the Open Science Framework and can be accessed using the following DOI: https://doi.org/10.17605/OSF.IO/AD6TR.