The metacognitive visual literacy framework for AI-generated explanations: an empirical study on the dual pathways to critical judgment accuracy

The proliferation of AI-generated visual explanations presents new challenges for visual literacy, particularly because such visuals may shape interpretation through algorithmic bias, omissions, or persuasive design choices. This study proposes and empirically tests a Metacognitive Visual Literacy (MVL) framework within Pakistan’s sociotechnical context. Drawing on survey data from 628 AI users and using partial least squares structural equation modelling, the study examines how AI-generated visual explanations are associated with users’ perceived visual critical judgement capacity. The findings show that this relationship operates through two distinct but parallel pathways: visual sense-making proficiency, which supports pattern recognition and coherent interpretation, and visual metacognitive awareness, which encourages reflection on personal assumptions, bias, and evaluative strategies. By positioning visual literacy as a critical, reflective, and culturally situated practice, the MVL framework extends visual literacy scholarship towards AI-mediated visual environments. The study contributes to current debates on visual literacy and AI explainability while also highlighting the need for future behavioural and experimental research to assess actual performance in evaluating AI-generated visuals.