From Outputs to Warranted Claims: Reframing AI Literacy as Culturally Mediated Quality Judgement for HRD

Generative AI has made the production of fluent drafts, summaries, and recommendations inexpensive and rapid, shifting the bottleneck of performance from producing outputs to judging whether outputs are good enough to use. This shift challenges prevailing accounts of AI literacy in human resource development (HRD) that emphasise tool proficiency and broad critical awareness while leaving the evaluative core under-specified. Drawing on an integrative literature review, this article reframes AI literacy as culturally mediated quality judgement: the capacity to interpret, appraise, justify, and responsibly act on AI-assisted outputs as warranted claims relative to domain standards and situational demands. The article makes three contributions: it (1) specifies four domains of quality judgement – relevance, warrant, applicability, and responsibility; (2) identifies the sociocultural conditions through which judgement is enacted and developed at work; and (3) outlines HRD design principles and testable propositions for cultivating calibrated reliance on AI assistance.