Evidence of What, for Whom? The Socially Contested Role of Algorithmic Bias in a Predictive Policing Tool

Overview

• This paper examines how different stakeholders in Chicago use evidence of algorithmic bias in a crime prediction tool to advance their own interests and agendas.
• The researchers conducted interviews with 18 community organizations, academics, and public sector actors to understand how they perceive and use the issue of algorithmic bias.
• The study finds that stakeholders employ diverse strategies to leverage algorithmic bias, from reforming policies to rejecting the tool, reframing crime as a structural issue, and challenging authority figures.
• The paper advocates for centering the perspectives and experiences of communities impacted by incarceration to ensure algorithmic bias can be used to challenge the status quo.

Plain English Explanation

This paper looks at how different groups in Chicago view and use the issue of algorithmic bias in a crime prediction tool. Algorithmic bias refers to when an algorithm produces unfair or discriminatory results, often reflecting biases present in the data or the way the algorithm was designed.

The researchers interviewed 18 people from community organizations, academic institutions, and the public sector to understand their perspectives on the crime prediction algorithm used by the Chicago police. They found that these stakeholders have very different ways of using the evidence of algorithmic bias.

Some use it to try to change the policies around how police patrol neighborhoods. Others completely reject the use of the algorithm-based policing tool. A few reframe the issue of crime as a larger, societal problem rather than just individual criminal acts. Some try to reveal information about authority figures in an effort to challenge their power. And some stakeholders, particularly those with more influence, use the algorithmic bias to reaffirm their own authority or existing power structures.

These varied uses reflect the long-standing tensions in criminal justice reform between the values of liberation and healing often emphasized by impacted communities, and the values of surveillance and deterrence that are often built into data-driven reform efforts.

The paper argues that to truly challenge the status quo, we need to center the perspectives and real-world experiences of the communities most affected by the criminal justice system, rather than just focusing on the technical details of algorithmic bias.

Technical Explanation

The researchers conducted a qualitative, critical study to understand how different stakeholders in the Chicago area perceive and use evidence of algorithmic bias in the context of the Chicago Police Department's crime prediction algorithm. They interviewed 18 participants from community organizations, academic institutions, and the public sector.

Through these interviews, the researchers identified several distinct ways that stakeholders strategically leverage evidence of algorithmic bias:

1. Reforming police patrol allocation policies
2. Rejecting algorithm-based policing interventions
3. Reframing crime as a structural rather than interpersonal problem
4. Revealing data on authority figures to subvert their power
5. Repairing and healing families and communities
6. Reaffirming the authority or existing power structures of more powerful actors

The researchers draw inspiration from Catherine D'Ignazio's taxonomy of refusing and using data to analyze these varied uses of algorithmic bias as evidence.

The paper argues that these different approaches reflect long-standing tensions in criminal justice reform between the values of liberation and healing often centered by system-impacted communities, and the values of surveillance and deterrence often instantiated in data-driven reform measures. The researchers advocate for centering the interests and experiential knowledge of communities impacted by incarceration to ensure that evidence of algorithmic bias can serve as a device to challenge the status quo.

Critical Analysis

The paper provides a nuanced and thoughtful analysis of how different stakeholders use the issue of algorithmic bias in the Chicago crime prediction tool to advance their own interests and agendas. The researchers acknowledge that these varied uses reflect deeper, long-standing tensions in the criminal justice reform landscape.

However, the paper does not delve into potential limitations or caveats of the research itself. For example, the relatively small sample size of 18 interviewees, while providing rich qualitative data, may not fully capture the breadth of perspectives on this issue within the Chicago area. Additionally, the paper does not explore potential biases or blind spots the researchers themselves may have brought to the analysis.

Further research could examine how the use of algorithmic bias evidence differs across other cities or regions, or how the strategies employed by stakeholders evolve over time. It would also be valuable to more deeply investigate the underlying values and assumptions that drive the divergent approaches to leveraging algorithmic bias highlighted in the study.

Overall, this paper makes an important contribution by illuminating the complex, politically-charged ways that evidence of algorithmic bias is deployed in the real-world context of criminal justice reform. However, continued critical examination of these dynamics will be crucial to ensuring that such evidence can truly be used to challenge systemic inequities.

Conclusion

This paper presents a nuanced analysis of how different stakeholders in Chicago strategically use evidence of algorithmic bias in a crime prediction tool to advance their own interests and agendas. The researchers find that stakeholders employ diverse approaches, from reforming policies to rejecting the tool, reframing crime as a structural issue, and challenging authority figures.

The paper argues that these varied uses reflect long-standing tensions in criminal justice reform between the values of liberation and healing often centered by impacted communities, and the values of surveillance and deterrence often instantiated in data-driven reform measures. The researchers advocate for centering the perspectives and experiences of communities affected by incarceration to ensure that evidence of algorithmic bias can truly be leveraged to challenge the status quo.

This study provides important insights into the real-world, political dynamics surrounding algorithmic bias in the criminal justice context. Continued critical examination of these dynamics will be crucial as we seek to ensure that evidence of algorithmic bias is used to drive meaningful, equitable change.