Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·Aug 7·5 min read
Negotiating Risk Boundaries in AI for Policing Through Mixed-Stakeholder Deliberation
TL;DR
A mixed-stakeholder workshop assessed 13 AI use cases in policing, focusing on racial bias. Participants were generally supportive of AI adoption, emphasizing the need for tools that work effectively and benefit all communities.
✦ Why It Matters
Engage with community stakeholders in AI discussions to ensure equitable outcomes in policing.
Key Takeaways
- Include community representatives in AI decision-making processes.
- Evaluate AI tools based on effectiveness and equitable benefits.
- Reject AI use cases that do not meet community needs.
- Adopt an inclusive design approach to risk-benefit analysis.
Full Summary
AI tools are increasingly used in policing, raising concerns about racial bias, particularly as affected communities are often excluded from decision-making. A workshop with 30 participants, including community representatives, police officers, and academics, evaluated 13 AI use cases, specifically examining their potential for racial bias.
While three use cases were outright rejected, including recidivism risk assessment, the overall sentiment was open to AI adoption. Discussions centered on critical questions regarding the effectiveness of these tools and their equitable benefits.
This approach, akin to the curb-cut effect in inclusive design, suggests that integrating racial equity considerations into AI risk-benefit analyses can enhance overall outcomes for all stakeholders.