The Impact of AI-Generated Hallucinations in Educational Settings: Trends, Gaps, and Future Directions

Abstract

The rapid integration of generative AI tools such as ChatGPT and Gemini into educational settings has transformed the way students access information, complete assignments, and engage with learning materials. These technologies offer efficiency and personalized support. However, they also pose risks, particularly in the form of ‘hallucinations,’ or factually incorrect content generated by AI. This study investigates how AI hallucinations affect student learning, trust in academic systems, and the ability to critically evaluate information. Using a systematic bibliometric approach, 193 peer-reviewed articles published between 2021 and 2026 were analyzed through Scopus, with visual network mapping performed in VOSviewer and trend analysis supported by Biblioshiny. The findings reveal that although interest in using AI and big data as a tool for knowledge discovery is growing, discussions around hallucinated content remain limited and fragmented. Co-occurrence analysis shows that terms like “ethics,” “misinformation,” and “trust” have only recently begun to gain traction, particularly after 2023. This paper identifies key trends, research gaps and future research including the absence of empirical evaluations of hallucination impact, the lack of detection frameworks, and insufficient AI literacy in curricula. It concludes with a call for future interdisciplinary efforts to develop robust safeguards, benchmark datasets, and policy frameworks to ensure the responsible and informed use of AI in learning environments.

Ähnliche Arbeiten