Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Articulation Work and Tinkering for Fairness in Machine Learning

Miriam Fahimi, Mayra Russo, Kristen M. Scott, Maria-Esther Vidal, Bettina Berendt, Katharina Kinder-Kurlanda

TL;DR

It is found that CS researchers engage with SOI research to some extent, but organizational conditions, articulation work, and ambiguities of the social world constrain the doability of SOI research for them.

Abstract

The field of fair AI aims to counter biased algorithms through computational modelling. However, it faces increasing criticism for perpetuating the use of overly technical and reductionist methods. As a result, novel approaches appear in the field to address more socially-oriented and interdisciplinary (SOI) perspectives on fair AI. In this paper, we take this dynamic as the starting point to study the tension between computer science (CS) and SOI research. By drawing on STS and CSCW theory, we position fair AI research as a matter of 'organizational alignment': what makes research 'doable' is the successful alignment of three levels of work organization (the social world, the laboratory, and the experiment). Based on qualitative interviews with CS researchers, we analyze the tasks, resources, and actors required for doable research in the case of fair AI. We find that CS researchers engage with SOI research to some extent, but organizational conditions, articulation work, and ambiguities of the social world constrain the doability of SOI research for them. Based on our findings, we identify and discuss problems for aligning CS and SOI as fair AI continues to evolve.

Articulation Work and Tinkering for Fairness in Machine Learning

TL;DR

It is found that CS researchers engage with SOI research to some extent, but organizational conditions, articulation work, and ambiguities of the social world constrain the doability of SOI research for them.

Abstract

The field of fair AI aims to counter biased algorithms through computational modelling. However, it faces increasing criticism for perpetuating the use of overly technical and reductionist methods. As a result, novel approaches appear in the field to address more socially-oriented and interdisciplinary (SOI) perspectives on fair AI. In this paper, we take this dynamic as the starting point to study the tension between computer science (CS) and SOI research. By drawing on STS and CSCW theory, we position fair AI research as a matter of 'organizational alignment': what makes research 'doable' is the successful alignment of three levels of work organization (the social world, the laboratory, and the experiment). Based on qualitative interviews with CS researchers, we analyze the tasks, resources, and actors required for doable research in the case of fair AI. We find that CS researchers engage with SOI research to some extent, but organizational conditions, articulation work, and ambiguities of the social world constrain the doability of SOI research for them. Based on our findings, we identify and discuss problems for aligning CS and SOI as fair AI continues to evolve.
Paper Structure (34 sections, 3 figures, 1 table)

This paper contains 34 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Contribution.
  3. Main findings.
  4. Structure of the paper.
  5. Background and related work
  6. Articulation work and the doability of research
  7. Facilitators and constraints.
  8. Tinkering.
  9. Method and Data Analysis
  10. Data collection and interview partners.
  11. Interview process and qualitative data coding.
  12. Research Ethics.
  13. Findings
  14. Social worlds: what does the fair AI world want?
  15. CS community.
  16. ...and 19 more sections

Figures (3)

  • Figure 1: Fujimura's three levels of work organization fujimura_constructing_1987, adapted for the case of fair AI. On the left side of the illustration, the three levels of work organization as per Fujimura's fujimura_constructing_1987 original conceptualization are depicted. To enable alignment, articulation work is needed, between levels. Further, the labeled lines (A & B) show how without alignment across all levels, doablity is not possible. On the right, we identify examples of actors and tasks that can be found across the three levels in line with our example.
  • Figure 2: Histogram Distribution of the h-index* for all 10 participants (*last five years).
  • Figure 3: Sankey Diagram. Represented are 450 conference and workshop papers from 2009 to 2023 from our interview partners to give an overview of publication venues and research sub-areas.