# Fiduciary Responsibility: Facilitating Public Trust in Automated Decision Making

**Authors:** Shannon B. Harper, Eric S. Weber  
**arXiv:** [2301.10001](https://arxiv.org/abs/2301.10001)  
**Date:** 2023-01-06  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2301.10001v1/

## Key points

- Lack of public trust in automated decision-making systems and deploying institutions
- Introduction of fiduciary responsibility within a data science lifecycle (DSL) as a solution
- Manifestation of bias in different stages of the DSL: data acquisition, model building, and post-processing
- Distinct mitigation strategies proposed for each stage
- Operating transparently and involving stakeholders throughout the DSL to build trust and accountability
- Case study on predictive policing technology used by the Los Angeles Police Department
- Actions that police departments could take to enhance fiduciary responsibility:
- Providing transparent communication through press releases and social media posts
- Holding town halls to address public concerns
- Utilizing multiple perspectives in decision-making processes
- Framework for promoting fairness, accountability, and transparency in automated decision-making systems to foster public trust

## Summary

This position paper addresses the lack of public trust in automated decision-making systems and the institutions that deploy them. It introduces the concept of fiduciary responsibility within a data science lifecycle (DSL) as a means to address this issue. The paper explores how bias can manifest in different stages of the DSL, including data acquisition, model building, and post-processing, and proposes distinct mitigation strategies for each stage. The authors argue that by operating transparently and involving stakeholders throughout the DSL, institutions can build trust and accountability in their decision-making processes. To illustrate these concepts, a case study on predictive policing technology used by the Los Angeles Police Department is examined. The paper concludes with specific examples of actions that police departments could take to enhance fiduciary responsibility, such as providing transparent communication through press releases and social media posts, holding town halls to address public concerns, and utilizing multiple perspectives in decision-making processes. Overall, this paper provides a framework for promoting fairness, accountability, and transparency in automated decision-making systems to foster public trust.