ChatGPT Usage and Attention Related Cognitive Errors in Daily Lives of University Students

Abstract

This research explored the relationship between the usage of ChatGPT and attention-related cognitive errors in undergraduates. Using cognitive load theory, dual process theory, and automation bias, the research examines how the various ways of using ChatGPT (i.e., academic writing aid, task support, and trust and reliance) influence attention lapses. A cross-sectional survey research design was adopted and data was collected from 385 university students with an age range of 18–24 years across various disciplines. The sample size was determined through G power analysis with medium effect size and 95% confidence interval. Standardized instruments such as the Attention-Related Cognitive Errors Scale (ARCES) and a validated ChatGPT Usage Scale were used. Correlation analysis determined that greater trust and reliance on ChatGPT was strongly associated with increased attention-related cognitive errors (r =.50, p <.01) but writing aid showed a non-significant relationship with ARCES (r = -.03). Whereas task support showed a positive relationship (r =.11, p <.05). Independent samples t test indicated female students showed higher attention related cognitive errors in contrast to males. Furthermore, students from humanities and social science disciplines had higher attention errors as compared to the ones from natural science and communication fields. Based on these findings, the artificial intelligence (AI) implementation in education reveals the complex cognitive impact, possibly hinting at its risks of excessive use. The research emphasizes the careful, responsible and ethical use of generative AI technologies such as ChatGPT to maintain some balance between convenience and cognitive engagement and development.

Ähnliche Arbeiten