Skip to main content

Strengthening Gender Equality in Generative AI

A report from Courtney Chau, AMWA’s young representative to the United Nations Department of Global Communication

On March 22, 2024, the United Nations (UN) Academic Impact organized a session titled “Strengthening Gender Equality in Generative AI” as part of the 68th Session of the Commission on the Status of Women – the UN’s largest annual gathering on gender equality and women empowerment. Goal #5 of the UN’s 2030 Sustainable Development Goals (SDGs) is to “Achieve gender equality and empower all women and girls.” Since the development of the 2030 SDGs in 2015, the AI market has grown substantially. Generative artificial intelligence (AI) produces content, such as text, images, or videos. While generative AI has brought innovation and entertainment, it also introduces threats that may exacerbate gender bias. The session “Strengthening Gender Equality in Generative AI” focused on highlighting the risks generative AI poses to gender equality and the potential to develop responsible, human-centered AI that reduces the gender divide. The panel featured several key women in the field of technology and AI: Dr. Revi Sterling, Dr. Kutoma Wakunuma, Dr. Jamiee Stuart, and Constanza Gomez Mont. 

The panelists highlighted the threat of exacerbating gender bias and amplifying regressive ideas about women and girls through generative AI. AI systems and technology development have largely been led and designed by men. According to a 2023 UNESCO Report, only 22% of professionals working in AI were women. Whether intentional or unintentional, the lack of women in the production of AI programs can lead to gender bias, not only in the datasets used to train algorithms, but in the leadership of companies who build these algorithms. Programmed biases can lead to discrimination, such as providing differential outputs based on gender. These biases can also manifest as stereotyping, for example, outputs promoting a stereotypical image of women that highlight domestic, sexualized, or less competent-appearing images compared to men. In fact, a 2024 UNESCO report of large language models discovered stereotypical gendered word associations. Female names were associated with the words “home”, “family”, and “child”, while male names were associated with “business”, “executive”, “salary”, and “career.” In addition to propagating regressive and misogynistic ideas, the biases programmed into generative AI can negatively impact the mental wellbeing of women.

Disparities also exist in the ability for women to access the benefits of AI, including learning experiences, personalized health and wellness information, and generation of business ideas. One of the targets of SDG #5 is to “Enhance the use of enabling technology, in particular information and communications technology, to promote the empowerment of women,” demonstrating that increasing access to technology is a critical component of closing the gender gap. According to the Women in the Digital Economy Fund, 1.5 billion women from low and middle income countries have no internet access and, in 2022, 259 million more men than women used the internet. Increasing women’s ability to access technology is essential to ensuring equitable benefit from AI.

The speakers highlighted opportunities to minimize gender bias and strengthen equality

in AI. In addition to diversity in datasets, diversity must be present in all levels of decision making in AI industries. The first step to addressing gender bias is increasing representation of women in leadership and creation of AI, making the voices of women part of the development and decision-making process. Gomez Mont noted that only 20% and 29% of women hold management positions in mid-market companies and enterprises, respectively. Within institutions, increasing representation can be achieved by allocating funds for gender parity plans and diversifying recruitment. In addition to recruitment, retention of women is equally important. According to Gomez Mont, due to challenges in work culture and microaggressions within the industry, 50% of women in tech roles leave these companies by the age of 35. Institutions can enable and retain women in the workforce with clear and strict harassment policies, leadership growth support programs, and reasonable maternity and paternity policies. At the level of governments and investors, investment in Science, technology, engineering, and mathematics (STEM) programs, Information and Communication Technologies (ICT) programs, and women’s entrepreneurship can encourage women to enter the technology workforce. Enhancing representation of women in the generative AI workforce sets the stage for creating equitable AI.

The second step to strengthening gender equality in generative AI is development of responsible AI that outsmarts gender discrimination. When designed and implemented thoughtfully and driven by social and ethical responsibility, AI can be inclusive in engaging and incorporating diverse voices. It is essential to center people over profit in order to create responsible, human-centered AI that aligns the values of developers and consumers. Institutions and governments can invest in research on the differential impact of AI on demographic groups, which can guide equitable AI development. By doing so, humanistic AI can prioritize positive human outcomes and benefit anyone, regardless of gender, by offering more accurate solutions, producing more innovation, promoting safe and peaceful ideas, and providing social and economic benefit to all. In this way, we can harness AI to promote equity.

Finally, AI regulatory policies to govern the deployment and development of AI should be enacted to protect citizens and minimize inequities. By 2030, the Southeast Asian economy is anticipated to gain $1 trillion through AI. Using the AI readiness index, the Southeast Asia region has been deemed to have the highest AI readiness, with Singapore being the most prepared to implement and harness the benefits of AI. A critical component of Southeast Asia’s preparedness to implement AI is the Association of Southeast Asian Nations (ASEAN) Guide on AI Governance and Ethics. This guide offers direction in considering the societal impact of AI to organizations seeking to design or deploy AI systems. AI can be implemented into many sectors, including energy, manufacturing, financial services, healthcare, and governmental programs. Their widespread applications make it even more essential to implement policies that ensure societal impacts of AI are at the forefront of development and implementation.

This session, “Strengthening Gender Equality in Generative AI,” highlighted the importance of creating human-centered AI. In the development of AI, while we must ensure gender parity, we must also recognize intersectionality in identities and develop AI that functions for the betterment of all individuals.

I am grateful to AMWA for the opportunity to participate in the 68th Session of the United Nations Commission on the Status of Women. As an organization affiliated with the UN Department of Global Communication,bAMWA is represented by leaders who participate throughout the year in meetings at the UN, in addition to the yearly CSW. I look forward to sharing more of those insights over the coming year.

AMWA Admin

Close Menu