Human-Centered AI in Practice: Peer Facilitators as Technostress Reducers in a Mixed-Methods LMS Study

Abstract

Purpose: While the rapid integration of generative artificial intelligence (AI) offers significant pedagogical potential, it often outpaces the psychological scaffolding that anxious learners require, creating a severe affective barrier known as technostress. To address this critical gap, this study investigates whether structured peer facilitation, embedded within a Learning Management System (LMS) based AI course, can effectively reduce undergraduate students' AI anxiety and strengthen their conceptual understanding. Methods/Study design/approach: An explanatory sequential mixed-methods design was conducted involving 33 undergraduate students encountering generative AI instruction for the first time. Quantitative data encompassing AI anxiety, conceptual understanding, and technical self-efficacy were collected pre- and post-intervention, followed by analysis using Wilcoxon signed-rank tests and Spearman's rank correlation. Subsequently, qualitative data from open-ended evaluations were examined through reflexive thematic analysis to contextualize and deeply explain the quantitative shifts. Result/Findings: The intervention yielded a statistically significant reduction in AI anxiety (Z = -2.151, p = .032) alongside a significant improvement in conceptual understanding (Z = -1.990, p = .047). Although technical self-efficacy and career projection improved, the changes did not reach statistical significance. Furthermore, facilitator evaluation exhibited a significant negative correlation with anxiety reduction (rs = -.389). Thematic analysis confirmed that peer facilitators effectively acted as technical translators and technostress reducers, establishing the psychological safety necessary for learning. Novelty/Originality/Value: This research uniquely repositions peer mediation as the vital affective "more capable other" within Vygotsky's Zone of Proximal Development, prioritizing the resolution of emotional barriers before cognitive engagement with AI tools. The study extends Human-Centered AI (HCAI) theory by delivering robust empirical evidence that pairing generative AI modules with human peer facilitation creates a synergistic learning ecosystem. This provides educators with a scalable, practical solution for mitigating technostress in modern digital education.