Dataset and Code for: The Productivity Paradox of AI-Assisted Software Development — Empirical Evidence, Longitudinal Quality Degradation, and a Formal Cost-Benefit Model

This dataset accompanies a manuscript submitted to MDPI AI. It contains: (1) CSV files with sensitivity analysis outputs for the formal cost-benefit model (break-even time t* and vulnerability density E[U]/N across parameter ranges); (2) extracted case study data from three published secondary sources (GitClear 2024, Pearce et al. 2022, Stack Overflow Developer Survey 2024); (3) the full model parameter table with central values, bounds, and empirical sources; (4) the governance framework specification; and (5) a fully self-contained Python script (Python 3.11, numpy, matplotlib, pandas) that reproduces all three manuscript figures and both sensitivity tables. No primary data was collected; all source data are from the original publications cited in the manuscript.