Predictive policing algorithms are racist. They need to be dismantled.

The article traces the origins of data-driven activism to a traumatic high school arrest, illustrating how historical police misconduct directly fuels systemic inequality. This personal experience motivated the co-founder of Data for Black Lives to expose how data is weaponized against Black communities, shifting the focus from street-level violence to the long-term consequences of algorithmic bias within the criminal justice system. Predictive policing tools, which forecast crime locations or individual reoffending risks, are shown to perpetuate racism through flawed inputs. Since arrest rates disproportionately target Black individuals, these algorithms learn from biased historical data, effectively automating discrimination. Even without explicit racial variables, socioeconomic proxies like zip codes and education levels act as substitutes for race, ensuring that marginalized groups remain over-policed and unfairly assessed by pretrial and sentencing tools. This narrative is critical to the open_data community as it highlights the dangers of unexamined, opaque algorithms in public safety. It underscores the urgent need for transparency, rigorous auditing, and inclusive data practices to prevent automated systems from reinforcing historical prejudices. Understanding these mechanisms allows data professionals to advocate for ethical standards that dismantle rather than digitize the school-to-prison pipeline.

Source: technologyreview.com
Published on 2023-12-13

Phi-2: The surprising power of small language models »

Open Data

Related news