FROM STREET-LEVEL TO EXCEPTION SINK: A SOCIO-TECHNICAL ANALYSIS OF DISCRETION, STRESS, AND INCONSISTENCY IN THE AUTOMATED BUREAUCRACY

The dataset contains 1,200 anonymized public service cases processed through a hybrid human AI decision system. It includes applicant vulnerability indicators, data quality and bureaucratic conditions, and measures of discretion, stress, and ethical conflict. Decision outcomes capture AI versus human handling, outlier cases, and exception routing. Overall, the data supports socio-technical analysis of inconsistency and pressure in automated bureaucracy.