Artificial Intelligence and Academic Integrity: Ethical Considerations for Graduate Students

Abstract

The explosion of generative artificial intelligence technologies has spurred a major paradigm shift that has significantly challenged and redefined key tenets of academic integrity, authorship and assessment in higher education. This research explores the ethical issues, usage of generative AI tools, and detection difficulties faced by graduate students working with G.A.I. The study uses a multi-methodological approach and data from a large sample of four hundred and fifty graduate students from a variety of disciplines to determine task-specific usage rates and discipline-specific differences in ethical acceptability. Moreover, the study compares the empirical effectiveness of commercial artificial intelligence detection systems by examining the receiver operating characteristic curve and related areas under the curve. The results show that artificial intelligence is widely used to refine structures and text and that it is not used at all to draft the theory. A distinct disciplinary divide can be seen, with students from STEM having a more liberal attitude towards the assistance provided from an algorithm, than students from the humanities. Most importantly, the study shows that detection software is fundamentally flawed, and can be misused in serious ways, giving rise to misaccusations. The study also reveals that a paradigm shift is necessary in higher educational institutions from reactive, surveillance-based policies to proactive, pedagogical ones aimed at the development of AI literacy, conceptualizing academic integrity within a collaborative postplagiarism environment.