Multidimensional Quality of Scientific Infographics Produced in a GenAI-Required Undergraduate Geology Assignment: A Single-Course Artifact-Level Study from Ecuador

Generative artificial intelligence (GenAI) is increasingly used to produce scientific–visual material, but a polished result is not evidence that its content is correct or its sources are checkable. This study examined 41 infographics on scanning electron microscopy (SEM) and its geological applications, produced in one undergraduate Geology course in Ecuador in an assignment requiring GenAI use. The unit of analysis was the infographic, not the student. Two evaluators scored every artifact with a study-specific eight-criterion rubric applied after submission. The mean weighted score was 72.93 ± 13.01 on a 25–100 scale. Scientific accuracy (mean 3.52 of 4) and visual hierarchy (3.38) were the strongest criteria, whereas source quality and traceability were the weakest (1.63), with most artifacts at the minimum score. Agreement on the total score was moderate but imprecise, and one criterion, verifiability of images and claims, was scored too inconsistently to support firm conclusions. Artifacts with clearer source evidence tended to integrate text and images more coherently. Within this single course, the results indicate which aspects of scientific material produced under a GenAI-required assignment require human checking. Sustainability is an implication of that verification work, a quality-assurance practice aligned with Sustainable Development Goal (SDG) 4, rather than an outcome measured here.