School of Languages at XJTLU Conference 2024

As the integration of artificial intelligence in educational settings becomes increasingly prevalent, students are found to utilize AI tools such as ChatGPT and Claud to different extents to complete their assignments. Some students heavily rely on AI tools to write a whole report without editing, while some students tend to use it more responsibly. At the same time, AI detectors are developed to spot plagiarism and unauthorized use of AI-generated content. However, these AI detector results are seldom trusted as a decisive indicator to determine students’ academic integrity in higher education. The percentage of AI writing advised by the tools may be treated as a reference, but the final decision is usually based on teachers’ judgment and students’ justification in investigation meetings. Does it suggest that AI detectors’ ability of identifying instances where human input is assisted by AI technology is not trustworthy? How do AI detectors use linguistic nuances, structural patterns, and semantic coherence to scrutinize written materials? Therefore, this presentation aims to evaluate the accuracy and reliability of AI detector results by comparing generated reports from the tool and students’ genuine use of AI in their assignments.