Atrition of Workers with Minoritized Identities on AI Teams Jefrey Brown ∗ San Diego State University jbrown4@sdsu.edu Tina Park Partnership on AI tina@partnershiponai.org Jiyoo Chang Partnership on AI jiyoo@partnershiponai.org Mckane Andrus University of Washington HCDE mkandrus@uw.edu Alice Xiang Sony AI alice.xiang@sony.com Christine Custis Partnership on AI christine@partnershiponai.org ABSTRACT The efects of AI systems are far-reaching and afect diverse commu- nities all over the world. The demographics of AI teams, however, do not refect this diversity. Instead, these teams, particularly at big tech companies, are dominated by Western, White, and male work- ers. Strategies for preventing harms done by AI must also include making these teams more representative of the diverse communi- ties that these technologies afect. The pipeline of students from K-12 and university level contributes to this - those with minori- tized identities are underrepresented or excluded from pursuing computer science careers. However there has been relatively little attention given to how the culture at tech companies, let alone AI teams, contribute to attrition of minoritized people in the workplace. The current study uses semi-structured interviews with minoritized workers on AI teams, managers of AI teams, and leaders working on diversity, equity, and inclusion (DEI) in the tech feld (N = 43), to investigate the reasons why these workers leave these AI teams. The themes from these interviews describe how the culture and climate of these teams may contribute to attrition of minoritized workers, and strategies for making these teams more inclusive and representative of the diverse communities afected by technologies developed by these AI teams. Specifcally, the current study found that AI teams in which minoritized workers thrive tend to foster a strong sense of interdisciplinary collaboration, support professional career development, and are run by diverse leaders who understand the importance of undoing the traditional White, Eurocentric, and male workplace norms. These go beyond the łquick fxesž that are prevalent in DEI practices. KEYWORDS Diversity, DEI, Racism, Sexism, Attrition ACM Reference Format: Jefrey Brown, Tina Park, Jiyoo Chang, Mckane Andrus, Alice Xiang, and Christine Custis. 2022. Attrition of Workers with Minoritized Iden- tities on AI Teams. In Equity and Access in Algorithms, Mechanisms, and ∗ Corresponding author. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for proft or commercial advantage and that copies bear this notice and the full citation on the frst page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specifc permission and/or a fee. Request permissions from permissions@acm.org. EAAMO ’22, October 06ś09, 2022, Arlington, VA, USA © 2022 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACM ISBN 978-1-4503-9477-2/22/10. . . $15.00 https://doi.org/10.1145/3551624.3555304 Optimization (EAAMO ’22), October 06ś09, 2022, Arlington, VA, USA. ACM, New York, NY, USA, 9 pages. https://doi.org/10.1145/3551624.3555304 1 INTRODUCTION AI teams have failed to refect the diverse communities their tech- nologies ultimately afect and, in some cases, harmed. Although AI as a feld has begun to reckon with the harms done to 1 minoritized or marginalized communities, 2020 saw an unprecedented increase in the number of organizations speaking out against racial justice. The murders of George Floyd, Breonna Taylor, Daunte Wright, and Ahmaud Arbery forced the United States and other countries all over the world to come to terms with the legacy of historic harms meted on entire communities of people based on race and intersections therein with other axes of identity such as ability, gender identity, and sexual orientation. In response to these mur- ders, subsequent public outcry, and demands from their employees, organizations released statements and pledged to increase the di- versity of their teams, among other eforts to show an attempt to pursue racial equity [18]. Despite this, AI teams have still refected the broader tech ecosystem in its makeup of mostly White, male workers [29]. Organizations have long focused on recruiting more diverse candidates for positions, especially those from marginalized or mi- noritized groups. However, relatively less efort has been focused on how the homogenous demographics of these teams have also infuenced minoritized individuals to leave these teams. The current paper will briefy describe the crucial need for improving the inclu- sivity of AI teams, present the results of an interview-based study with minoritized individuals on AI teams to ask them why they have left or continue to stay on AI teams, and propose recommendations for making these teams more inclusive. 2 RATIONALE The rationale to study the attrition of minoritized workers in the AI feld is three-fold. First, the harms associated with AI are dis- proportionately borne by historically minoritized communities. Buolamwini and Gebru’s seminal work Gender Shades powerfully demonstrated how existing societal biases can be encoded in algo- rithms, in this case, bias in classifying the faces of Black women [4]. Several researchers have since shown how bias can be encoded in other domains including hiring, mortgage approval, and approval for credit lines [23, 24]. Second, the people working on AI do not 1 This paper uses the term minoritized, coined by Gutnaratnam [15] to emphasize the active minoritization or marginalization process meted upon groups with less institutional power than the dominant groups. This is elaborated upon in the methods section. Terms such as łminorityž does not accurately capture this.