Academic Dishonesty In The Age Of Artificial Intelligence: The Role Of University Students' Epistemological Beliefs

Abstract

Digital technologies and generative artificial intelligence have reshaped university learning by expanding access to information while also creating new pathways for academic misconduct. This study examined university students' epistemological belief profiles, their engagement in internet-triggered academic dishonesty, perceived contributing factors, and the extent to which belief dimensions predict dishonest behavior in online contexts. A descriptive-correlational design was used to obtain data from 279 students at a public state university in the northern Philippines using the Internet-Triggered Academic Dishonesty Scale and the Epistemological Beliefs Questionnaire. Data were analyzed using descriptive statistics, Spearman's rank-order correlation, and multiple linear regression. Findings revealed generally low to modest engagement in internet-facilitated dishonesty, with delinquency recording the highest composite mean and fraudulence the lowest, suggesting that routine, collaborative, and citation-related lapses were more evident than deliberate deception. Students also showed developing epistemological beliefs, particularly in active knowledge acquisition and valuing learning, although beliefs about learning speed and ability remained less mature. Overall, epistemological beliefs were significantly and negatively correlated with internet-triggered academic dishonesty (rho = -0.196, p = 0.001). Regression results further revealed that acquisition of knowledge (beta = -0.255, p = 0.002) and speed of learning (beta = -0.244, p = 0.007) significantly predicted dishonest behavior, with the model explaining 13.3% of the variance. The study concludes that academic integrity in the digital age requires not only sanctions but also educative policies, responsible AI guidance, process-based assessment, and programs that cultivate epistemic maturity, persistence, and ethical ownership of learning.