A tridimensional model of AI literacy: An empirical analysis of student performance and demographic patterns in higher education

Abstract

This study moves beyond theoretical frameworks to empirically analyse artificial intelligence (AI) literacy among undergraduate students, identifying distinct performance profiles to inform educational interventions. Using a validated, performance-based instrument, we assessed the functional, technical and socio-critical competencies of 353 students at a private university in Mexico. Our analysis revealed three distinct student profiles: lower performance (n = 85), mid-range proficiency (n = 158) and higher competence (n = 107). A critical finding across all profiles was a significant deficit in the socio-critical dimension, with only 2.6% of students demonstrating outstanding ability. Furthermore, the profiles varied significantly by gender and academic stage, challenging traditional assumptions about technology literacy. These findings provide an evidence-based typology for diagnosing student needs and developing targeted, equitable educational strategies to foster comprehensive AI literacy in higher education.

Implications for practice or policy:

• Curriculum leaders should view AI literacy as a transversal competence, integrating ethical and critical reflection across curricula.
• Educators and instructional designers must apply differentiated instruction based on learner profiles, recognising that AI literacy development is complex and challenges assumptions about gender and academic progression.
• Policymakers should promote validated assessment tools to replace anecdotal evidence with empirical data, guiding institutional strategies and resource allocation for AI education.