Integrating STEAM education with ethical frameworks (SDGs 4, 5, and 10, and Global Competence) imposes a critical cognitive load on teachers, limiting their efficiency and instructional agency. While Artificial Intelligence (AI) promises interactive evaluative scaffolding, its reliability demands pedagogical transparency. This study validates a STEAM pedagogical rubric – comprising 20 indicators and 60 descriptors evaluated by 14 experts – conceived as a preliminary blueprint for outcome-directed ethical constraints in future AI-assisted evaluation systems. Quantitative results demonstrated highly satisfactory content validity (S-CVI ≥.85), establishing a stable semantic baseline with strong human-led consensus (Kendall’s W = 0.714, p < .001). Crucially, thematic analysis revealed a tension between theoretical exhaustiveness and practical feasibility, identifying cognitive overload and time constraints as significant barriers to manual implementation. This operational bottleneck highlights the potential utility of digital scaffolding to support teacher well-being. By establishing a human-validated pedagogical baseline, this research offers a structured reference framework designed to support the ethical alignment of automated assessment with global sustainability agendas. Ultimately, the framework serves as an initial stepping stone toward mitigating algorithmic opacity and anchoring future AI-driven designs under human pedagogical oversight.
Sagastizabal-Sáez et al. (Mon,) studied this question.