Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Making Transparency Advocates: An Educational Approach Towards Better Algorithmic Transparency in Practice

Andrew Bell, Julia Stoyanovich

TL;DR

This work addresses the gap between XAI research and real-world transparency by testing an educational approach that creates transparency advocates within organizations. It introduces the Algorithmic Transparency Workshop, an open-source, domain-tailored training program designed to increase literacy and motivate advocacy across news/media and startup domains. Through interviews and pre/post surveys with 27 professionals, the study finds that the workshop enhances understanding of algorithmic transparency and stimulates actions at three levels: conversational, implementational, and influential, with domain-specific barriers shaping outcomes. The findings suggest a viable path to translate responsible AI practices into practice by empowering ground-up change, and the authors share all workshop materials for broad adoption and replication.

Abstract

Concerns about the risks and harms posed by artificial intelligence (AI) have resulted in significant study into algorithmic transparency, giving rise to a sub-field known as Explainable AI (XAI). Unfortunately, despite a decade of development in XAI, an existential challenge remains: progress in research has not been fully translated into the actual implementation of algorithmic transparency by organizations. In this work, we test an approach for addressing the challenge by creating transparency advocates, or motivated individuals within organizations who drive a ground-up cultural shift towards improved algorithmic transparency. Over several years, we created an open-source educational workshop on algorithmic transparency and advocacy. We delivered the workshop to professionals across two separate domains to improve their algorithmic transparency literacy and willingness to advocate for change. In the weeks following the workshop, participants applied what they learned, such as speaking up for algorithmic transparency at an organization-wide AI strategy meeting. We also make two broader observations: first, advocacy is not a monolith and can be broken down into different levels. Second, individuals' willingness for advocacy is affected by their professional field. For example, news and media professionals may be more likely to advocate for algorithmic transparency than those working at technology start-ups.

Making Transparency Advocates: An Educational Approach Towards Better Algorithmic Transparency in Practice

TL;DR

This work addresses the gap between XAI research and real-world transparency by testing an educational approach that creates transparency advocates within organizations. It introduces the Algorithmic Transparency Workshop, an open-source, domain-tailored training program designed to increase literacy and motivate advocacy across news/media and startup domains. Through interviews and pre/post surveys with 27 professionals, the study finds that the workshop enhances understanding of algorithmic transparency and stimulates actions at three levels: conversational, implementational, and influential, with domain-specific barriers shaping outcomes. The findings suggest a viable path to translate responsible AI practices into practice by empowering ground-up change, and the authors share all workshop materials for broad adoption and replication.

Abstract

Concerns about the risks and harms posed by artificial intelligence (AI) have resulted in significant study into algorithmic transparency, giving rise to a sub-field known as Explainable AI (XAI). Unfortunately, despite a decade of development in XAI, an existential challenge remains: progress in research has not been fully translated into the actual implementation of algorithmic transparency by organizations. In this work, we test an approach for addressing the challenge by creating transparency advocates, or motivated individuals within organizations who drive a ground-up cultural shift towards improved algorithmic transparency. Over several years, we created an open-source educational workshop on algorithmic transparency and advocacy. We delivered the workshop to professionals across two separate domains to improve their algorithmic transparency literacy and willingness to advocate for change. In the weeks following the workshop, participants applied what they learned, such as speaking up for algorithmic transparency at an organization-wide AI strategy meeting. We also make two broader observations: first, advocacy is not a monolith and can be broken down into different levels. Second, individuals' willingness for advocacy is affected by their professional field. For example, news and media professionals may be more likely to advocate for algorithmic transparency than those working at technology start-ups.

Paper Structure

This paper contains 35 sections, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Organizational barriers to transparency.
  4. Regulation.
  5. Tempered radicals.
  6. AI education.
  7. Methods
  8. The Algorithmic Transparency Workshop
  9. Development process.
  10. Structure and design.
  11. Breakout activity.
  12. Recruitment, Participation, and Domains of Study
  13. Content customization.
  14. Data Collection and Analysis
  15. Interviews.
  16. ...and 20 more sections

Figures (2)

  • Figure 1: Completed breakout room activity from one of the four breakout rooms with media professionals. From top-to-bottom, the green cards (reasons for transparency) read: "I'd like to know if humans or robots are reviewing my comments. I expect to be talking to humans.", "People blame us for decisions. We need to explain them.", "It can build trust with our users.", "Easier to evaluate", "We want credit for the work", "Open source builds community and support." From top-to-bottom, the red cards (reasons against transparency) read: "It's taking our jobs!", "Helps out our competitors", "It's more work, work we could be spending improving the product itself." Illustrations by Falaah Arif Khaan.
  • Figure :