Algorithmic Policing
Algorithmic policing is the practice of using computer algorithms and Machine Learning to assist police departments with crime prevention. These algorithms include Predictive Policing and Algorithmic Surveillance technologies.
Predictive Policing Technologies are used to predict where and by whom crimes will be committed in the future.
- E.g. The VPD uses GeoDASH to predict the areas in Vancouver which are most likely to experience break-and-enters within a given time period.
Algorithmic Surveillance Technologies assist police with monitoring and surveillance of citizens. These technologies included license-plate readers, social media surveillance tools, and face-recognition software.
Note: The Electronic Frontier Foundation maintains a database of police surveillance technology in use across the United States at https://atlasofsurveillance.org/.
Law enforcement agencies across Canada have begun to use algorithmic policing software. However, many people are concerned about the ethical implications of these technologies.
This table shows the algorithmic policing technologies in use by police departments in Canada as of Sept. 2020. (Source)
Ethical implications
The neural networks used by these tools are generally trained on historical data, leading to Algorithmic Bias.
These technologies may create positive Feedback Loops: a pattern of crime is noticed so police pay special attention to that sector, leading to more arrests and a more evident pattern.