Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

On the Societal Impact of Machine Learning

Joachim Baumann

TL;DR

The thesis tackles the societal impact of ML by establishing measurement frameworks, decomposing bias dynamics, and designing interventions to improve fairness without sacrificing utility. It links domain-specific fairness metrics to normative theories, provides a structured decomposition of ML pipelines and their feedback loops, and develops practical interventions including optimal decision rules, algorithmic collective action, and social-good deployments such as proactive rental assistance. Eight papers across theory and applied domains supply a cohesive, empirically grounded framework for measuring, anticipating, and mitigating algorithmic discrimination. The work emphasizes continuous challenges and future directions, including the integration of generative AI and participatory approaches to ensure ML aligns with broader social values and equity goals.

Abstract

This PhD thesis investigates the societal impact of machine learning (ML). ML increasingly informs consequential decisions and recommendations, significantly affecting many aspects of our lives. As these data-driven systems are often developed without explicit fairness considerations, they carry the risk of discriminatory effects. The contributions in this thesis enable more appropriate measurement of fairness in ML systems, systematic decomposition of ML systems to anticipate bias dynamics, and effective interventions that reduce algorithmic discrimination while maintaining system utility. I conclude by discussing ongoing challenges and future research directions as ML systems, including generative artificial intelligence, become increasingly integrated into society. This work offers a foundation for ensuring that ML's societal impact aligns with broader social values.

On the Societal Impact of Machine Learning

TL;DR

The thesis tackles the societal impact of ML by establishing measurement frameworks, decomposing bias dynamics, and designing interventions to improve fairness without sacrificing utility. It links domain-specific fairness metrics to normative theories, provides a structured decomposition of ML pipelines and their feedback loops, and develops practical interventions including optimal decision rules, algorithmic collective action, and social-good deployments such as proactive rental assistance. Eight papers across theory and applied domains supply a cohesive, empirically grounded framework for measuring, anticipating, and mitigating algorithmic discrimination. The work emphasizes continuous challenges and future directions, including the integration of generative AI and participatory approaches to ensure ML aligns with broader social values and equity goals.

Abstract

This PhD thesis investigates the societal impact of machine learning (ML). ML increasingly informs consequential decisions and recommendations, significantly affecting many aspects of our lives. As these data-driven systems are often developed without explicit fairness considerations, they carry the risk of discriminatory effects. The contributions in this thesis enable more appropriate measurement of fairness in ML systems, systematic decomposition of ML systems to anticipate bias dynamics, and effective interventions that reduce algorithmic discrimination while maintaining system utility. I conclude by discussing ongoing challenges and future research directions as ML systems, including generative artificial intelligence, become increasingly integrated into society. This work offers a foundation for ensuring that ML's societal impact aligns with broader social values.

Paper Structure

This paper contains 38 sections, 1 figure.

Table of Contents

  1. Introduction
  2. Background
  3. Automated decision-making and recommender systems.
  4. Real-world cases of algorithmic discrimination.
  5. Group fairness criteria in ML.
  6. Dynamic aspects of fairness in ML systems.
  7. The multidisciplinary nature of ML's societal impact.
  8. Three Goals Towards Fairness in ML
  9. Measure
  10. What's in a metric?
  11. Measuring societal impact.
  12. Decompose
  13. Beyond symptom treatment.
  14. Bias can enter at multiple stages.
  15. Static analysis misses dynamic effects.
  16. ...and 23 more sections

Figures (1)

  • Figure 1: Summary of key ideas (I), methods (M), findings (F), and contributions (C) for the eight papers included in this thesis, organized by the three main research goals.