Caribbean lecturers’ navigation and perceptions of academic integrity and AI-detection software in English for academic purposes (EAP) modules

Abstract

The rapid proliferation of Generative Artificial Intelligence (GenAI) has fundamentally reshaped academic writing practices in higher education, prompting renewed debates surrounding authorship, academic integrity, ethical AI use, and assessment validity. These challenges are particularly pronounced in English for Academic Purposes (EAP) contexts, where language production constitutes a central learning outcome. This study investigates Caribbean EAP lecturers' perceptions of academic integrity in the GenAI era, focusing on evolving understandings of integrity, concerns regarding AI-detection technologies, strategies for balancing learning support and integrity enforcement, and professional development needs. Guided by Virtue Ethics and Gilligan's (1982) Ethics of Care, the study adopted a qualitative phenomenological approach. Data were collected through a mixed-methods questionnaire administered to 18 EAP lecturers across three Caribbean territories. Quantitative data were analyzed using descriptive statistics, while qualitative responses were subjected to content analysis. Findings indicate that lecturers increasingly conceptualize academic integrity as encompassing transparency, accountability, ethical AI use, and meaningful human contribution rather than merely the absence of plagiarism. Lecturers expressed significant scepticism towards AI-detection software, citing concerns about false positives, reliability, and potential biases against multilingual and Creolophone learners. However, rather than advocating prohibition, lecturers favored pedagogical approaches centered on assessment redesign, reflective practice, AI literacy, and responsible AI integration. However, participants also reported inadequate institutional guidance and limited professional preparation for addressing AI-related integrity challenges. The study proposes the Transparent Human-Centred AI Integrity (THAI) Model, which positions transparency, human agency, ethical responsibility, AI literacy, and institutional support as key pillars for navigating academic integrity in AI-mediated multilingual learning environments. As one of the first studies to examine these issues from a Caribbean EAP perspective, the research contributes a contextually grounded understanding of how educators negotiate the opportunities and risks of GenAI and offers practical implications for policy development, professional learning, and assessment design in higher education.