<![CDATA[This qualitative case study examines how English lecturers exercise professional judgment when generative artificial intelligence (GenAI) is introduced into student writing assessment and feedback. Ten English lecturers at Universitas Khairun participated, producing 30 documented judgment episodes across three student writing samples. Each episode linked an initial lecturer assessment, AI-generated information from ChatGPT, Claude, or Gemini, and a final judgment. Data from assessment records, follow-up interviews, student texts, AI-generated outputs, and contextual documents were examined through within-episode and cross-episode thematic analysis. The findings show that AI did not move lecturers' scores in a single direction. Rather, it prompted them to retain, lower, or raise initial judgments after reconsidering the evidence. Lecturers evaluated AI output in relation to construct alignment, evidentiary weight, learner needs, task purpose, local context, and feedback usability. Professional agency was evident in how they determined which parts of the AI output could influence the final assessment. The study describes this process as selective evaluative mediation and argues that GenAI introduces a second object of professional judgment: lecturers judge not only student performance but also the relevance, usability, and authority of AI-generated interpretations of that performance.]]>
Copyrights © 2026
