Reimagining writing assessment for the AI era: a systematic review on balancing AI support and authentic skill growth

This systematic review synthesizes empirical evidence from 19 studies examining AI (AI) integration in academic writing assessment across higher education contexts. Following PRISMA 2020 guidelines, the review addresses five interconnected research questions regarding AI tool functionality, stakeholder perceptions, student motivations, institutional implementation gaps, and assessment redesign strategies. Findings reveal that generative AI platforms, particularly ChatGPT (88.8% adoption) and Grammarly (67.4%), dominate across writing stages, with students strategically deploying tools based on cognitive complexity levels. Stakeholder perspectives diverge significantly: students value productivity enhancement and accessibility, faculty prioritize academic integrity concerns, and administrators adopt cautious openness with disproportionate resource allocation favoring faculty guidance (67%) over direct student support (17.8%). Student motivations for inappropriate use stem from structural vulnerabilities including time pressures, ethical ambiguity, and institutional policy voids rather than malicious intent. Geographic inequities persist, with Global South institutions facing infrastructure limitations that amplify educational disparities. Promising pedagogical and assessment innovations include process-oriented assessment, oral defenses, multimodal composing, and explicit AI literacy curricula-each targeting the redesign of writing evaluation practices to balance AI assistance with authentic skill development. The review concludes that effective integration requires moving beyond detection-focused approaches toward comprehensive pedagogical redesign grounded in equity, authentic learning, and development of uniquely human capacities for critical judgment and ethical reasoning.