@article{ART003380390},
author={Kang Sun},
title={Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -},
journal={Legal Theory & Practice Review},
issn={2288-1840},
year={2026},
volume={14},
number={3},
pages={195-216}
TY - JOUR
AU - Kang Sun
TI - Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -
JO - Legal Theory & Practice Review
PY - 2026
VL - 14
IS - 3
PB - The Korea Society for Legal Theory and Practice Inc.
SP - 195
EP - 216
SN - 2288-1840
AB - Predictive policing is a technology that enhances the efficiency of police operations by using artificial intelligence to predict areas or targets with a high likelihood of crime. However, this technology faces limitations in securing democratic legitimacy and public trust due to issues such as algorithmic opacity, bias, and unclear accountability. Recently, Explainable Artificial Intelligence (XAI) has been proposed as a solution to these issues; however, current XAI focuses solely on technical explanations of the premises underlying the results derived by the algorithm, while failing to include the selection process of the input data and variables that form the basis of those results within the scope of its explanations. Focusing on these limitations, this study presents a theoretical rationale for expanding the scope of explainability from the algorithm’s inference stage to the input stage. To this end, by examining Tim Miller’s concept of explanation, Foucault’s concepts of power-knowledge and temporal governmentality, and Habermas’s concept of democratic discourse, this study argues that the selection of data and variables used in predictive policing is not merely a matter of technical design but a process involving social and political value judgments, and that it must therefore be subject to democratic oversight and social legitimacy. Based on this, this study proposes the concept of “input explainability,” which extends the scope of XAI to the input stage.........
KW - Predictive policing;explainability;XAI;input explainability;democratic legitimacy;temporal governmentality
DO -
UR -
ER -
Kang Sun. (2026). Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -. Legal Theory & Practice Review, 14(3), 195-216.
Kang Sun. 2026, "Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -", Legal Theory & Practice Review, vol.14, no.3 pp.195-216.
Kang Sun "Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -" Legal Theory & Practice Review 14.3 pp.195-216 (2026) : 195.
Kang Sun. Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -. 2026; 14(3), 195-216.
Kang Sun. "Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -" Legal Theory & Practice Review 14, no.3 (2026) : 195-216.
Kang Sun. Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -. Legal Theory & Practice Review, 14(3), 195-216.
Kang Sun. Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -. Legal Theory & Practice Review. 2026; 14(3) 195-216.
Kang Sun. Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -. 2026; 14(3), 195-216.
Kang Sun. "Reconstructing Explainability in Predictive Policing - From Outcome-Oriented XAI to Democratic Control of Input Data -" Legal Theory & Practice Review 14, no.3 (2026) : 195-216.
