Artificial Intelligence (AI) Risk Management Framework

Workforce diversity, equity, inclusion, and accessibility processes are prioritized in the mapping, measuring, and managing of AI risks throughout the lifecycle.

Governance and Oversight, AI Design

A diverse team that includes AI actors with diversity of experience, disciplines, and backgrounds to enhance organizational capacity and capability for anticipating risks is better equipped to carry out risk management. Consultation with external personnel may be necessary when internal teams lack a diverse range of lived experiences or disciplinary expertise.

To extend the benefits of diversity, equity, and inclusion to both the users and AI actors, it is recommended that teams are composed of a diverse group of individuals who reflect a range of backgrounds, perspectives and expertise.

Without commitment from senior leadership, beneficial aspects of team diversity and inclusion can be overridden by unstated organizational incentives that inadvertently conflict with the broader values of a diverse workforce.

Organizational management can:

- Define policies and hiring practices at the outset that promote interdisciplinary roles, competencies, skills, and capacity for AI efforts.
- Define policies and hiring practices that lead to demographic and domain expertise diversity; empower staff with necessary resources and support, and facilitate the contribution of staff feedback and concerns without fear of reprisal.
- Establish policies that facilitate inclusivity and the integration of new insights into existing practice.
- Seek external expertise to supplement organizational diversity, equity, inclusion, and accessibility where internal expertise is lacking.
- Establish policies that incentivize AI actors to collaborate with existing nondiscrimination, accessibility and accommodation, and human resource functions, employee resource group (ERGs), and diversity, equity, inclusion, and accessibility (DEIA) initiatives.

Organizations can document the following
- Are the relevant staff dealing with AI systems properly trained to interpret AI model output and decisions as well as to detect and manage bias in data?
- Entities include diverse perspectives from technical and non-technical communities throughout the AI life cycle to anticipate and mitigate unintended consequences including potential bias and discrimination.
- Stakeholder involvement: Include diverse perspectives from a community of stakeholders throughout the AI life cycle to mitigate risks.
- Strategies to incorporate diverse perspectives include establishing collaborative processes and multidisciplinary teams that involve subject matter experts in data science, software development, civil liberties, privacy and security, legal counsel, and risk management.
- To what extent are the established procedures effective in mitigating bias, inequity, and other concerns resulting from the system?

AI Transparency Resources
- WEF Model AI Governance Framework Assessment 2020. [URL](https://www.pdpc.gov.sg/-/media/Files/PDPC/PDF-Files/Resource-for-Organisation/AI/SGModelAIGovFramework2.pdf)
- Datasheets for Datasets. [URL](http://arxiv.org/abs/1803.09010)

Diversity, Interdisciplinarity, Governance

Dylan Walsh, “How can human-centered AI fight bias in machines and people?” MIT Sloan Mgmt. Rev., 2021. [URL](https://mitsloan.mit.edu/ideas-made-to-matter/how-can-human-centered-ai-fight-bias-machines-and-people)

Michael Li, “To Build Less-Biased AI, Hire a More Diverse Team,” Harvard Bus. Rev., 2020. [URL](https://hbr.org/2020/10/to-build-less-biased-ai-hire-a-more-diverse-team)

Bo Cowgill et al., “Biased Programmers? Or Biased Data? A Field Experiment in Operationalizing AI Ethics,” 2020. [URL](https://arxiv.org/pdf/2012.02394.pdf)

Naomi Ellemers, Floortje Rink, “Diversity in work groups,” Current opinion in psychology, vol. 11, pp. 49–53, 2016.

Katrin Talke, Søren Salomo, Alexander Kock, “Top management team diversity and strategic innovation orientation: The relationship and consequences for innovativeness and performance,” Journal of Product Innovation Management, vol. 28, pp. 819–832, 2011.

Sarah Myers West, Meredith Whittaker, and Kate Crawford,, “Discriminating Systems: Gender, Race, and Power in AI,” AI Now Institute, Tech. Rep., 2019. [URL](https://ainowinstitute.org/discriminatingsystems.pdf)

Sina Fazelpour, Maria De-Arteaga, Diversity in sociotechnical machine learning systems. Big Data & Society. January 2022. doi:10.1177/20539517221082027

Mary L. Cummings and Songpo Li, 2021a. Sources of subjectivity in machine learning models. ACM Journal of Data and Information Quality, 13(2), 1–9

“Staffing for Equitable AI: Roles & Responsibilities,” Partnership on Employment & Accessible Technology (PEAT, peatworks.org). Accessed Jan. 6, 2023. [URL](https://www.peatworks.org/ai-disability-inclusion-toolkit/ai-disability-inclusion-resources/staffing-for-equitable-ai-roles-responsibilities/)