#936: How Students' Mental Models of AI Shape Their Epistemic Cognition of AI-Generated Results: Source and Justification

The introduction of generative artificial intelligence (AI) has changed students’ learning environments. Such rapid changes have the potential to exacerbate the post-truth challenges we face today. Despite the rapid increase in the use of AI in education, there remains a lack of research on how students reason about AI-generated results. Aligned with models of epistemic cognition, this study investigated students’ mental models about AI and how such models influence their epistemic cognition about AI, particularly regarding source and justification. Results revealed that students’ epistemic reasoning varies depending on their mental models of AI. Students with a web-scraping model indicated that AI-generated results could be unreliable and superficially justified due to the source of information (i.e., the Internet). Students with a training-dataset model emphasized human intention in selecting training data and highlighted critical flaws in AI justification processes based on their understanding of how AI systems function.