Gender and functional differentiation in generative AI usage among Malaysian higher education student

Despite achieving gender parity in higher education enrolment (60% female), Malaysia faces an emerging digital divide in how students use artificial intelligence. This study examined whether gender predicts task-specific ChatGPT usage patterns among Malaysian students (n = 443), employing latent profile analysis and multinomial logistic regression on CC BY 4.0 licensed Global ChatGPT Student Survey data (October 2024–February 2025). Four distinct usage profiles emerged: Selective Users (14.9%), Moderate Adopters (31.8%, coding-focused), Academic Enthusiasts (33.0%, text-focused), and Comprehensive Users (20.3%). Gender significantly predicted specialized profile membership (χ² = 19.47, p < .001). Males concentrated in coding-focused use (OR = 0.48 for females, p = .007), females in text-focused use (OR = 1.89, p = .020), while Comprehensive Users exhibited gender parity. Exploratory analyses indicate a tentative pattern of larger gender gaps in technical AI use in STEM fields (33.4 percentage points in Applied Sciences) versus Social Sciences (12.7 points), though small cell sizes (n=3 for female Natural Sciences students) prevent definitive conclusions. Profiles predicted domain-specific skill development with large effects (η² = .18-.33). Findings reveal that equal access masks unequal functional engagement, with implications for gender-segregated occupational pathways that warrant further investigation with larger samples.