The Wyndham Grand hotel at Downtown Pittsburgh was filled with 500+ brilliant minds between October 5-7, who are exploring issues related to educational measurement. The conference AIME-Con 2026 organized by the National Council of Measurement in Education (NCME) gave opportunities to these scholars to present their groundbreaking works that could potentially take AI and measurement to the next level.
Ahnaf presented his co-authored paper, “Who Keeps Up with Whom? Assessing GenAI Text Detection Across Version Updates,” challenging the reliability of commercial AI detection tools in an era of rapidly evolving generative AI. Drawing on an empirical analysis of 960 text samples, the study examined how chatbot and detector updates, AI humanization techniques, and writing styles influence detection accuracy. The findings suggest that generative AI chatbots are advancing faster than detection technologies can reliably keep pace, raising concerns about the accuracy of widely used tools such as Turnitin and GPTZero. The research highlights the potential consequences of unreliable AI detection for academic integrity, particularly the risk of misidentifying student work. Beyond identifying these limitations, the study advocates for evidence-based institutional policies and improved detection strategies to promote fair, responsible, and trustworthy AI integration in education.
Ahnaf presenting the work on-behalf of the co-authors at AIME-Con 2026. (Photo: Golnoush Haddadian)
Distinguished keynote speakers of the conference included Alina A. von Davier, Chief of Assessment at Duolingo, and Dr. Dragan Gašević, Professor and Director of the Centre for Learning Analytics at Monash University.
To know more about the work, please refer to this post.
Leave a Reply