The Disparity Of Black Women In The AI Industry

 

The Disparity Of Black Women In The AI Industry

Introduction

The AI industry, like many other tech sectors, faces significant challenges regarding diversity and inclusion. Black women, in particular, are vastly underrepresented, which has far-reaching implications for both the industry and society at large.

Underrepresentation: Black women make up a small fraction of the tech workforce. For instance, in the UK, Black women constitute only 0.7% of those working in technology. In major tech companies in the US, less than 2% of employees in technical roles are Black.

Bias in AI Systems: We often assume that machines are impartial, but they are not. My research has revealed significant gender and racial biases in AI systems developed by major tech companies like IBM, Microsoft, and Amazon. When tasked with identifying the gender of a face, these systems performed much better on male faces than female faces. For lighter-skinned men, the error rates were below 1%, while for darker-skinned women, the error rates skyrocketed to 35%. AI systems from these leading companies have even failed to correctly identify the faces of prominent figures such as Oprah Winfrey, Michelle Obama, and Serena Williams. When technology misidentifies even these iconic women, it is crucial to re-evaluate how these systems are designed and who they are truly serving.

The underrepresentation of Black women in technology is a multifaceted issue rooted in historical, social, and economic barriers. Here are some key factors that have contributed to this disparity:

Historical Barriers

Educational Inequities: Historically, Black women have faced significant barriers to accessing quality education. Segregation and underfunded schools have limited opportunities for many Black students, particularly in STEM fields. The events of 2020 drew renewed attention to longstanding inequalities in the invention and innovation ecosystem and Black Americans’ complex relationship with technology. In the United States, Black men and women are less likely than other demographics to earn STEM degrees, receive patents, or commercialize new products and services. Black scientists and engineers experience unconscious bias and outright discrimination in the high-tech employment sector, while Black inventor-entrepreneurs face persistent difficulties in gaining access to venture capital, intellectual property protection, and commercial networks. With Black technologists largely absent from the invention process, supposedly neutral apps and algorithms are encoded with racist assumptions that perpetuate negative stereotypes and deepen social inequality. And yet, Black Americans regularly invent, tweak, and deploy technology in the course of cultural and political expression and develop new products and services with global reach.

Economic Disparities: Economic inequities have also played a crucial role. The digital divide was first recognized in the mid-1990s.1 Three decades later, due in part to long-standing economic inequity and the economics of broadband, it remains an impediment to inclusive economic growth, particularly in Black American communities. Approximately 40 percent of Black American households—as opposed to 28 percent of White American households—don’t have high-speed, fixed broadband.  In dense urban areas such as Chicago and Baltimore, Black households are twice as likely as their White counterparts to lack a high-speed internet subscription. In the rural South, 38 percent of Black households don’t have broadband, compared with 23 percent of White households.

Stereotypes and Bias: Persistent stereotypes, such as the “angry Black woman” trope, and systemic biases have created hostile environments in both educational and professional settings. Lower levels of digital readiness are both a consequence and an ongoing driver of large gaps in Black American representation in jobs that require digital skill sets. Although Black Americans comprise approximately 13 percent of all workers, they make up only 7.4 percent of digital workers.

The stereotypes and biases faced by Black women in AI technology are significant and multifaceted. Here are some key aspects:

Racial and Gender Bias in AI Systems: Persistent stereotypes, such as the “angry Black woman” trope, and systemic biases have created hostile environments in both educational and professional settings. Lower levels of AI digital readiness are both a consequence and an ongoing driver of large gaps in Black women representation in jobs that require digital skill sets.  For example, facial recognition technologies have been shown to have higher error rates for darker-skinned individuals, particularly women. This can lead to misidentification and exclusion from various services and opportunities.

1. Stereotypes in AI Outputs: AI-generated content can perpetuate harmful stereotypes. For instance, image generators and other AI tools have been found to produce results that reinforce racial and gender stereotypes, such as associating certain professions or characteristics with specific races or genders.
2. Workplace Discrimination: Black women in AI often face discrimination and bias in the workplace. This includes being overlooked for promotions, receiving less support from colleagues, and experiencing microaggressions. These challenges can create a hostile work environment and hinder career progression.

Lack of Representation and Mentorship: The absence of Black women in AI tech leadership roles means fewer role models and mentors for young Black girls entering the field. This lack of representation can perpetuate feelings of isolation and imposter syndrome. Many Black women face challenges in accessing quality education in STEM fields. Schools in predominantly Black communities are often underfunded, which limits access to advanced courses and resources necessary for pursuing AI careers.

1.     Intersectionality of Race and Gender: The unique challenges faced by Black women arise from the intersection of racial and gender biases. This intersectionality means they often experience compounded discrimination that can hinder their career progression moving forward.  Black women face unique challenges due to the intersection of racial and gender biases. This means they can experience compounded discrimination that is not simply the sum of racial and gender biases but a unique form of bias that affects them differently.

Representation in AI Development: The lack of diversity in AI development teams can exacerbate these biases. When AI systems are developed without considering the diverse experiences of different groups, the resulting technologies can perpetuate existing inequalities.

Conclusion

Diversity in AI is crucial because it ensures that the technology we develop is inclusive, fair, and representative of all users. When diverse perspectives are included in the creation of AI systems, it helps to mitigate biases and prevents the perpetuation of existing inequalities.

Without diversity, AI systems can unintentionally reinforce stereotypes and exclude marginalized groups, leading to unfair outcomes. For example, facial recognition technology has been shown to have higher error rates for people with darker skin tones due to a lack of diverse training data.

Moreover, a diverse workforce in AI fosters innovation and creativity, as different viewpoints and experiences contribute to more robust and effective solutions. By prioritizing diversity, we can build AI systems that better serve society as a whole and drive positive social change.

Excerpts from 

Joy Buolamwini's article Artificial Intelligence Has a Problem With Gender and Racial Bias. Here’s How to Solve It