Artificial Intelligence Hallucination Risk Assessment Using the SOAR Model: How Possible is it?

Abstract

The integration of artificial intelligence (AI) into educational practices has become increasingly prevalent, particularly among secondary school students who utilize this technology to access information and complete academic tasks. While artificial intelligence offers considerable advantages in educational contexts, it also introduces potential risks, notably the phenomenon of 'AI hallucinations,' wherein the system generates outputs that appear credible but are, in fact, inaccurate, irrelevant, or entirely fabricated. This phenomenon can undermine students’ digital literacy and critical thinking. This study aims to develop a valid and reliable risk assessment scale for AI hallucinations, based on four key constructs which is factual, contextual, multimodal, and logical hallucinations. The development of this scale is guided by the SOAR Model (Strengths, Opportunities, Aspirations, Results), which emphasizes a strategic, strengths-based, and future-oriented approach. The key findings indicate that the SOAR model effectively demonstrates its potential by identifying individual strengths, challenges, available options, responses, and overall effectiveness within an educational context. The scope was deliberately confined to the SOAR model to encourage constructive engagement with the development of AI hallucination mitigation strategies. It also contributes to the development of a more comprehensive and ethical AI literacy curriculum for stakeholder’s purposes. The implications can benefit educational policy, teacher professional development, and improvements in AI systems. Future research could expand upon this analysis by employing alternative futuristic frameworks to gain broader perspectives and greater analytical depth. The scale is designed not only as a diagnostic tool but also as a foundation for targeted pedagogical interventions. It can be used by teachers and counsellors to identify students at high risk and to plan appropriate intervention programs. Overall, the initiative supports efforts to build a generation of students who are technologically literate, critical thinkers, and capable of using AI responsibly.

Ähnliche Arbeiten