Inteligencia Artificial en la predicción delictiva: Análisis crítico de su eficacia y sesgos
Revista Logos Ciencia & Tecnología · 29.12.2025 · 18 (1)
DOI: 10.22335/rlct.v18i1.2234 ↗Kurz gefasst
Fachartikel in der Zeitschrift Revista Logos Ciencia & Tecnología, online erschienen am 29.12.2025. Verfasst von Nicolas Marchal González. Die Arbeit wurde bisher 1-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
This article provides a comprehensive analysis of the application of Artificial Intelligence (AI) in crime prediction. It traces the evolution from the theoretical foundations of environmental criminology and managerial systems like CompStat to modern deep learning architectures (e.g., CNN-LSTM, ST-ResNet) designed for spatio-temporal analysis. The paper critically evaluates the documented efficacy of these systems, contrasting precision successes in controlled settings with the mixed results and failures in real-world implementations (e.g., PredPol, HunchLab). The central thesis argues that predictive AI implementation is inherently compromised by ethical challenges, primarily biased input data ("dirty data") and self-reinforcing feedback loops that amplify racial and social disparities. Finally, it explores technical solutions (Explainable AI - XAI) and, crucially, democratic governance frameworks (like the EU AI Act and CCOPS) as indispensable requirements for any responsible deployment
Zitieren
APA
González, N. M. (2025). Inteligencia Artificial en la predicción delictiva: Análisis crítico de su eficacia y sesgos. Revista Logos Ciencia & Tecnología, 18(1). https://doi.org/10.22335/rlct.v18i1.2234
Vancouver
González NM. Inteligencia Artificial en la predicción delictiva: Análisis crítico de su eficacia y sesgos. Revista Logos Ciencia & Tecnología. 2025;18(1). doi:10.22335/rlct.v18i1.2234
BibTeX
@article{Gonzlez2025,
author = {González, Nicolas Marchal},
title = {Inteligencia Artificial en la predicción delictiva: Análisis crítico de su eficacia y sesgos},
journal = {Revista Logos Ciencia & Tecnología},
year = {2025},
volume = {18},
number = {1},
doi = {10.22335/rlct.v18i1.2234}
}
EUVIMED