Summary and the Path Forward
Summary and the Path Forward
This study used interviews with minoritized workers on AI teams, managers of AI teams, and DEI leaders to investigate 3 main questions about workers in AI:
- Why do minoritized folks leave AI teams?
- What influence does the culture of these teams have on whether these workers stay or leave?
- What can be done to make these teams more inclusive?
Approaching these questions using interviews and qualitative analysis created the potential to discover in-depth themes and ask follow-up questions about a topic that has yet to be studied with such specificity. The themes that emerged demonstrated the importance of a strong alignment of an organization’s values with the values of the workers on AI teams. Teams which fostered a strong sense of interdisciplinary collaboration, and respect for diverse professional and personal identities, tended to attract and retain diverse individuals. Managers and senior leadership were crucial in maintaining the organizational policies necessary for these teams to thrive. They tended to emphasize the importance of DEI beyond tokenism or superficial trainings mandated by HR.
No easy solution exists to transform an AI team from homogeneous and toxic to diverse and inclusive. Systemic change is difficult and often goes against a company’s profit motive.
Teams must fundamentally question their norms and values, something which White and middle-class tech workers from prestigious institutions are not used to doing.
However, this is necessary to truly make room for diverse perspectives which are necessary to avoid further harm to minoritized populations around the globe.
Since the summer of 2020, companies have redoubled both their internal and public-facing commitments to DEI. The interviews revealed that several of their efforts have been effective, such as supporting ERGs. These are not enough to transform systemically racist or sexist cultures and can in fact backfire if taken over by leaders who do not understand the fundamental mission of DEI.
This interview study is just a starting point from which much more work needs to be done to determine the nuances behind building more inclusive workplaces. In particular, studies specific to more local contexts can better understand some of the specific nuances behind DEI all over the world. The limitations of the study must also be acknowledged. Considering a probable selection bias, the study cannot make claims or generalize towards the experiences of all minoritized individuals on AI teams. Rather, the study gives deep insight into common themes that emerged across a selection of workers who opted into the study.