Predictive policing software can lend false objectivity to existing police bias

Evidence

The program involves reliance on a software system— PredPol—that instantiates an algorithm that examined ten years worth of crime data, including the types of crimes committed, dates and times, and locations. The software produces a prediction as to where and when crimes are likely to occur over the succeeding twelve hours. … It has been alleged by critics that the software provides a patina of objectivity to what is mere bias: officers want to over-police predominantly Black neighborhoods, and the software gives them an explicit reason to do so (LA Times Editorial Board, 2019). — Joseph Avery, Jim Sidanius, Joel Cooper, Gregory Davis, Amanda N., Bias in the Law, 1161

Categorization

• XREF: The ‘patina of objectivity’ critique connects directly to the COMPAS passages in earlier chapters — algorithmic legitimacy laundering bias. — connects to: Palantir’s software predicts individual violence likelihood for police — Shoshana Zuboff, The Age of Surveillance Capitalism, 1236
• SEED: The feedback-loop problem — police data feeds the algorithm which directs policing which produces more data — is a strong angle for a piece on why predictive policing entrenches, rather than corrects, existing bias.