Abstract
Generative artificial intelligence (AI) tools are now widely accessible to the public. This popularization has brought about new academic integrity challenges in higher education. Meanwhile, there is still no clear conclusion on how college teachers form their judgments of students’ assignments completed with the help of AI tools. This study focuses on cognitive biases. These biases systematically influence teachers’ ability to distinguish students’ original writing from AI-generated writing. This research is guided by dual-process theory and the heuristic-bias framework. It identifies four major types of cognitive biases among educators. The four categories include automation bias and algorithm aversion, anchoring effects, the representativeness heuristic, and confirmation bias. This paper adopts conceptual analysis methods and sorts out the latest relevant research findings. It explores how the above-mentioned cognitive biases lead to two common evaluation errors in teaching. The two errors are false accusations of student academic misconduct and failure to identify AI-assisted writing. Moreover, these errors produce uneven impacts on different groups of students. The study holds that the assessment model that centers on AI detection has inherent flaws. Such limitations stem from the cognitive characteristics of human evaluators. Therefore, this paper suggests that educators should transform the assessment mode and adopt a process-oriented assessment design. This study also puts forward specific practical suggestions. These include using standard-ized evaluation scoring criteria, recording and assessing students’ whole learning process, and building a campus cultural atmosphere focused on critical AI literacy. Combined with the above research analysis, this paper concludes that aca-demic integrity construction in higher education should not rely merely on supervision and management. Instead, it should be regarded as an educational practice. This educational practice needs to fully adapt to the actual cognitive rules of teaching evaluators.
References
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2024). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 61(2), 228-239.
Cummings, M. L. (2004). Automation bias in intelligent time critical decision support systems. AIAA 1st Intelligent Systems Technical Conference, 2004-6313.
Evans, J. S. B. T., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science, 8(3), 223-241.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7), 100779.
Morewedge, C. K., Yoon, H., Scopelliti, I., Symborski, C. W., Korris, J. H., & Kassam, K. S. (2015). Debiasing decisions: Improved decision making with a single training intervention. Policy Insights from the Behavioral and Brain Sciences, 2(1), 129-140.
Perkins, M. (2023). Academic integrity considerations of AI large language models in the post-pandemic era: ChatGPT and beyond. Journal of University Teaching & Learning Practice, 20(2), 07.
Sadasivan, V. S., Kumar, A., Balasubramanian, S., Wang, W., & Feizi, S. (2023). Can AI-generated text be reliably detected? arXiv preprint arXiv:2303.11156.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131.