Improving students' openness to artificial intelligence through risk awareness and digital literacy: Evidence form a developing country

Abstract

The integration of technology, particularly artificial intelligence (AI) in learning process has the potential to enhance learning efficiency by offering diverse perspectives and quick information access. However, this reliance on AI can lead to overdependence, thereby diminishing the overall learning experience. This study aims to explore student acceptance and risk factors for learning using AI through the Technology Acceptance Model (TAM). This cross-sectional study surveyed 218 accounting students using a structured questionnaire. The internal consistency of latent constructs was verified through Cronbach's alpha, followed by exploratory factor analysis to ensure the unidimensionality of these constructs. The theoretical framework was tested using Partial Least Squares - Structural Equation Modeling (PLS-SEM). The result suggests that students with higher self-confidence demonstrate greater enthusiasm for learning AI, while digital literacy significantly influences the perceived ease of use. While accessibility is not prioritized, functionality and information accuracy are deemed more critical. The research model posits that perceived usefulness, perceived ease of use, self-efficacy, digital literacy, and perceived risk impact student acceptance of AI. Perceived usefulness and self-efficacy positively influence student acceptance, while perceived risk has a negative impact. Furthermore, it emphasizes the importance of increasing basic technological literacy among students in online learning environments. Despite these insights, the study is limited by its focus on accounting students, and future research should consider a broader demographic. This study contributes to the existing literature by highlighting the role of digital literacy and self-efficacy in AI adoption, offering valuable implications for educators and AI service providers.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 24 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.