AX-5R Replication Materials: Ablation Discriminant Probe Data and Code

Data and analysis code for the supplementary ablation discriminant probe reported in the manuscript "From AI Adoption to AI Transformation: The AX-5R Framework for Socio-Technical Work-System Redesign" (2026). Two cross-provider language models (gpt-4o-mini and claude-haiku-4-5-20251001) generated 252 implementation artifacts across three workflows and seven prompt arms (the full AX-5R frame, five single-component ablations, and a sham five-part control), then scored by independent blinded cross-provider judges on five transformation-grade criteria. The package contains the generated artifacts, independent judge scores, the reported analysis output, analysis scripts, and documentation. The reported results (Table 5 effect sizes and inter-rater reliability r = 0.66) are reproducible from the included data without API keys. Results are descriptive evidence of rating separation, not confirmatory or field evidence.