From Frustration to Fluency: Prior Familiarity, Cognitive Load, and Digital Equity in AI-Assisted LaTeX Authoring Among Secondary Mathematics Students

Despite growing interest in AI-assisted writing tools, little research has examined how prior technological familiarity shapes the adoption experience for secondary school students learning professional mathematical typesetting. This study investigates the relationship between prior familiarity, cognitive load dimensions, and technology acceptance when 45 secondary mathematics students (aged 16-18) used Overleaf, a collaborative LaTeX editor with AI code-completion features, for the first time. Drawing on the Technology Acceptance Model (TAM) and Sweller's Cognitive Load Theory (CLT), we examine how six NASA-TLX workload dimensions differentially relate to prior experience, and whether demographic factors (device type, previous digital tool) moderate the adoption experience. Results showed good overall usability (SUS M = 72.94, α = .91) and low-to-moderate cognitive load (TLX M = 32.03), with prior familiarity significantly predicting reduced frustration (rs = -.45, p_adj = .013) and effort (rs = -.36, p_adj = .047) after Benjamini-Hochberg correction. A hierarchical regression model incorporating all six TLX dimensions explained 39.8% of variance in acceptance (p = .006). Crucially, no significant differences emerged between device types or previous digital tools, supporting a digital equity interpretation: the AI-assisted LaTeX environment was equally accessible regardless of students' technological backgrounds. These findings contribute to understanding how AI-augmented authoring tools can democratise access to professional mathematical communication in secondary education.