Abstract

Generative AI tools promise to democratise programming education by lowering barriers to entry for underrepresented learners. Yet nascent research suggests a more complex phenomenon. This research-in-progress paper examines how generative AI reconfigures the pedagogical infrastructure of programming education through an in-depth qualitative investigation of a UK-based social enterprise training disadvantaged individuals for technology careers. Drawing on 35 interviews, 84 hours of observational data, and document analysis, we reveal how AI integration fundamentally reshapes learning arrangements, not by simplifying them, but by redistributing and intensifying labour. We theorise this transformation as a shift to “fragmented apprenticeship”, where epistemic authority fragments across algorithmic outputs, human validators, and learners themselves. Three key findings emerge: AI functions as augmented learning infrastructure requiring continuous human calibration; learners must develop boundary work competencies to navigate legitimate AI use; and significant invisible labour emerges as volunteers engage in constant mediation and validation. Paradoxically, technologies positioned as equalisers may instead reproduce inequality by imposing new labour costs that different groups are differently equipped to sustain. Our findings challenge techno-optimistic narratives and offer preliminary design implications for genuinely inclusive AI-mediated education.

Download

Share

COinS