This project seeks to investigate the ethical implications of the use of Automated Decision Systems (ADS) in government administration and their subsequent impact on public trust in democratic institutions. We are seeing that artificial intelligence is being increasingly deployed across high stakes public sectors such as calculating criminal recidivism scores for bail, directing police resources for predictive mapping, and automating the allocation of welfare benefits. As these procedures transfer from the discretion of human agents to the decision making of complex algorithms, the core machinery of public governance is becoming detached from human explanation.
The central research question driving this study is: What is the relationship between the use of automated decision making in government and the public's level of trust in democratic accountability? While proponents of ADS argue that algorithmic decision making enhances institutional efficiency and eliminated human bias, critics suggest that these black-box systems compromise due process.
Using a mixed-methods approach this research blends political philosophy with empirical social science. This is achieved through deploying a scenario based national survey experiment to measure the tension between technological efficiency and human agency. This study aims to evaluate the sustainability of the democratic social contract in a world in which the rules governing citizens are increasingly being hidden behind complex code.
The primary objective of this project is to evaluate how the deployment of opaque automated decision making systems by government agencies influence public trust and the perceived legitimacy of democratic institutions. Secondary objectives include analyzing whether it is the black box nature of algorithms that impact public trust or the general idea of automated technology itself, as well as analyzing whether the public prioritizes the Hobbesian idea of institutional efficiency or the Rawlsian demand for public justification from state power.
This project sits at the intersection of political philosophy, data ethics, and public policy. The theoretical foundation of this project is drawn from the the classic idea of social contract as well as the democratic "publicity principle", which dictates that the rules governing a society must be known, transparent, and capable of being publicly justified to all citizens. When the state outsources its decision making functions to algorithms, it creates what other scholars have referred to as a "black box of delegated authority." Existing literature on data ethics such as Virginia Eubanks' Automating Inequality and Cathy O'Neil's Weapons of Math Destruction heavily inform this project and establish the ideas that automated tools profile and punish historically marginalized communities under the false idea of reduced biases. This research aims to bridge a cap in the existing literature by providing empirical numbers on how automated systems alter civic behavior and institutional legitimacy. As local and federal governments rapidly increase AI powered automation, the findings of this project will hopefully provide a humanistic lens to a field focused on efficiency. The data gathered can help inform future policymakers in creating AI regulations and frameworks to protect the foundational integrity of the democratic social contract and protect vulnerable populations.
As I proceed with this project, I would love to connect with anyone in the Laidlaw Network who has experience with empirical social science research methods. I would incredibly appreciate feedback or guidance on the statistical analysis of my survey data. Additionally, if anyone has connections to public administrators or policy advocates working directly with these types of automated systems, I would greatly appreciate an introduction.
Thank you for reading through my research outline! I would love to hear any thoughts, questions, or critiques.