An exploratory study of examining the effects of graduate-level learners’ trust towards GAI-enhanced 3D virtual teachers on learning outcomes, behavioral transitions, and perceived cognitive load

Abstract

Recently, generative AI (GAI) has aroused great attentions in education. However, whether learners’ trust concerning GAI would hinder self-regulated learning remain understudied. This study examines the effects of graduate-level learners’ trust concerning GAI-enhanced 3D virtual teachers on their learning outcomes, behavioral transitions, and perceived cognitive load. Six graduate-level learners were recruited in a research-intensive university in Hong Kong, and they were grouped into high-level trust group (HT group), medium-level trust group (MT group), and low-level trust group (LT group) according to their beforehand self-reported data, respectively. Datasets were collected from final concept map products, process-oriented data, and survey data. We used a mixed approach to analyze these data. Results showed that the HT group exhibited the best learning outcomes, the most reasonable behavioral transitions, and appropriate perceived cognitive load in their learning. While learners in the LT group performed worse, the MT group’s performance was average. This findings provide valuable insights for educators when integrated GAI in learning, guiding instructors and learners to use artificial intelligence to smooth their learning.