Digital submission platforms, automated writing evaluation, and generative artificial intelligence (GenAI) have unsettled the evidential basis on which English writing lecturers judge student texts. Research on AI-assisted writing, automated feedback, and teacher beliefs is expanding, yet it has rarely been synthesised around the cognitive conflict lecturers experience when their professional judgment collides with digital evidence and institutional expectations. This systematic literature review, reported in line with PRISMA 2020, examined how English writing lecturers experience and manage such conflict, what generates it, and which strategies they use to resolve it. Twenty-six full-text reports were assessed for eligibility, and 19 empirical studies published between 2014 and 2026 in 13 countries were analysed through thematic synthesis. Lecturers experienced conflict chiefly as epistemic uncertainty about authorship, as divergence between holistic judgment and machine output, and as an affective and identity burden marked by anxiety, guilt, fatigue, and a drift from assessor to “detective.” Institutional and pedagogical sources appeared in all 19 studies, technological sources in 18, and ethical and student-related sources in 16 each; policy vacuums and unreliable detection were the most decisive triggers. Lecturers relied mainly on verification (16 studies) and communicative strategies (14 studies), whereas institutional responses were largely recommended rather than enacted (13 versus 5 studies). Many conflicts were therefore deferred or displaced rather than resolved. The review proposes a source–discrepancy–management–outcome model and argues that fair digital writing assessment requires process-visible task design, explicit AI-use disclosure, collegial moderation, and an expanded assessment literacy that includes the interpretation of algorithmic evidence.
Copyrights © 2026