Rethinking Assessment in Accounting Education amid Generative AI Dwi Fitra Arreski, Fadhilah Ahmad Qaniah
Universitas Terbuka, Universitas Negeri Gorontalo
Abstract
Abstract
Generative artificial intelligence is reshaping the validity, integrity, and purpose of assessment in accounting education. This study presents a systematic literature review of 15 peer-reviewed studies on generative AI, assessment, and accounting education. The review examines how GenAI affects existing accounting assessments, what redesign strategies are proposed in the literature, and which competencies should be prioritized in AI-mediated learning environments. Four main themes were identified. First, GenAI is increasingly capable of completing routine and structured accounting assessment tasks, although its performance remains less reliable in higher-order tasks that require judgement and contextual reasoning. Second, academic integrity risks cannot be addressed effectively through AI detection alone, especially in unsupervised and text-based assessment. Third, the literature points toward authentic, process-oriented, and higher-order assessment designs, including supervised applied tasks, oral defense, longitudinal projects, and AI-integrated critique activities. Fourth, accounting programs need to assess AI literacy, ethical reasoning, evaluative judgement, professional skepticism, and communication, not only procedural correctness. The review concludes that accounting educators should move beyond narrow anti-cheating responses and redesign assessment systems that make student reasoning, verification, and professional judgement more visible.
