<![CDATA[This study explores the perceptions of English lecturers toward GPT-5 as an Automated Essay Scoring (AES) tool, specifically focusing on its accuracy in content-related evaluation within Indonesian higher education. A qualitative approach was employed, using convenience sampling to select three lecturers at State Islamic University Syarif Hidayatullah Jakarta for semi-structured interviews. Data were analyzed using thematic analysis. Findings indicate that ChatGPT effectively recognizes linguistic features and content coherence, making it a capable co-rater when prompts and rubrics are clearly defined. However, lecturers expressed "partial trust," noting that the AI may miss narrative nuances and stylistic depth. Similar studies support this finding by pointing out the AI’s swift development that enhances its ability in evaluating contents, especially when the instructions/prompts, rubrics, and any supporting information are clear. This research suggests that, though ChatGPT could be used as AES, its implementation should be under human control due to possible error and ethical responsibility.]]>
Copyrights © 2025
