Artificial intelligence has transformed the emerging technologies space, creating revolutionizing impact in all spheres of life. Can gender-sensitive policies be introduced in the domain of artificial intelligence?
Participation and engagement of different voices are necessary when novel technologies are adopted for solving societal problems. Lack of proper representation is the reason behind the skewed nature of the landscapes and systems we see today. As Artificial Intelligence is gaining momentum as a tool for economic and digital transformation, we need to be cautious about the possibility of processes being locked according to the views of one side. AI has potential, but how it is governed shapes the impacts it has on society. However, the policy and ethics institutions influencing the AI field suffer from a lack of diversity, especially in terms of gender. Women remain starkly underrepresented in prominent AI policy organizations, meaning female perspectives are often overlooked. This omission can lead to biased, unethical AI that amplifies gender discrimination. Greater inclusion of women and their viewpoints in AI governance is crucial to create policies that consider all members of society.
Contextualizing the problem
According to a study by the World Economic Forum in 2021, the analytics and related AI positions have only 26% women participation. Now, after Sam Altman’s rehiring, the newly elected director board of OpenAI is dominated by men. What this news implies is that the historical biases that existed in the tech industry are being carried over into leadership roles. This scenario is not ideal as AI is spreading with boundless energy across the globe. There is also a growing demand for AI/ML-oriented skills recently, but women as a talent pool remained an unexplored idea. A possible solution for this is effective policies addressing the needs of markets and society. But the reality is that the AI policymaking space also has a very low female engagement. The reasons are obvious and visible, but remain unaddressed.
The first reason is education. There are fewer women studying and working in STEM fields relevant to AI development, like computer science, engineering, and mathematics. This has been an ongoing discussion for so many years and proposed as a reason for various gender-related problems in various fields. But the gap is not getting filled as it should be. It is evident from the mere 4% increase in women doing PhDs in AI and CS in North America in a ten-year interval (2010-19)(Women in AI, 2019). The underrepresentation of women in AI begins early, as social stereotypes and lack of exposure dissuade girls from pursuing STEM education necessary for AI careers.
Also, women in STEM education face isolation as minorities and lack guidance on pathways into the ambiguous AI field. All of these will cause low growth in the talent pipeline which leads to fewer women in AI research and development, which in turn will influence the percentage of AI governance participation.
AI policymaking requires expertise not just in technology but also laws, ethics, and policy. If fewer women enter tech fields, a smaller pool of women experts exists to move into policy roles later.
Then, the age-old story of biases in opportunities. Most women lack recruitment, internships, and career guidance in AI while students, hence missing key chances to enter the field. Then the implicit biases in AI recruiting and overemphasis on engineering/technological terms limit women's entry, as many lack awareness of alternative AI career paths. These unconscious (or conscious) gender biases may lead to fewer women being recruited, hired, or promoted into AI policymaking roles. In addition, implicit biases cause women to face extra pressure in work settings. According to a Deloitte report on ‘Women in AI’, females have to continuously prove that they have enough experience to work on a specific task while the expertise of a male colleague is not questioned. This creates a less welcoming environment for women in the tech industry, which can also be the case for leadership and policymaking roles. They also lack visible female role models and face bias when they don't fit the male archetype of leadership. There are reports of high levels of gender discrimination, lack of recognition, and poor management from the women employed in the field of AI. In a Deloitte survey, it was revealed that over half of women leave AI jobs due to this treatment, contributing to high attrition of women in the field. There are also workplace obstacles like lack of flexible schedules, parental leave, and support for caregivers which may disproportionately affect women's ability to advance in this space.
The potential consequences
The quality of data is what makes an AI system work, but what it actually shows is the perspectives of the developers. Most of the videos, texts and pictures used to train an AI system are dominated by the history of men. These biases along with ongoing systemic gaps women face, will lead to the narrative of AI being shaped primarily by men. This is evident in the recent conversations about AI, which centered on profits and competition while the societal impact is receiving very little space. This can be reduced if there is diversity among the people overseeing the process of development of the AI systems. So, policy-making and governance require female participation which will help in detecting the biases and other concerns perpetuated by the system early on but the lack of women's perspectives in policy leaves blind spots.
We have seen examples of the lack of diverse oversight enabling problematic AI applications that disproportionately harm women. Like the resume screening software from Amazon exhibiting bias against female applicants. Algorithms can’t be blamed in this instance because the training data was full of ‘male’ candidates from Amazon in technical positions. Fortunately, the software was dropped after discovering the bias, but having women in the development team of the software is very important to anticipate the effect earlier. Similarly, Google’s medical AI system was found to underperform for breast cancer due to limited data. Such failures demonstrate the real-world dangers of sidelining women’s perspectives in AI development and governance. If data represents history, due to the discriminations and sidelining decades ago, it will not be a reliable asset for making these applications. When the people designing them can easily bypass the regulations because the policies do not have the right principles, technology will become a nightmare for the users.
Important ethical and social considerations may be overlooked if policymaking is dominated by one demographic's priorities. Women's concerns and their needs won’t be addressed in the policies. For example, if an application is designed for hygiene education, there is information and problems that can only be detected and explained by women. A team of men overseeing the standards of such a project won’t be able to find anything wrong with the application if it doesn’t have any data regarding periods, but a woman can. Also, women can have different perspectives regarding ethics and values, which will make the policies more inclusive. Women and marginalized groups could be put at risk of harm from AI systems that embed societal biases. Tech-facilitated Gender-Based Violence is increasing with the arrival of many technologies. Without policies that can properly understand and define the problems and solutions, this scenario is going to worsen in the coming years.
Another important side effect of this is the public trust in AI could erode if systems seem unfair or biased against women, and policy seems unresponsive to a large part of society. This happens in many countries where continuous policy failures cause citizens to lose interest in governance mechanisms. If AI policymaking is also biased against half of the population, that can affect the relationship between people and organizations. Also, the economic potential of AI technologies could be limited if large swaths of talent are left out of their development. And women may be discouraged from entering tech fields if policymaking seems exclusionary, limiting diversity even further and the future generations will have fewer role models.
Addressing the question
Despite proven benefits of diversity, women are vastly underrepresented in AI. Many face discrimination driving high attrition. Though challenges persist, businesses can take steps to increase women's representation, retention, and leadership in AI. Achieving gender equity requires fundamentally valuing women's equal contributions. Companies and governments can actively recruit and appoint more women to leadership and policy roles in AI ethics boards, regulatory bodies, and advisory councils. An inclusive culture which transparently measures and addresses gender gaps in AI, while actively working to eliminate bias and promote diversity is the key.
Provide mentoring, training, and opportunities to help prepare women and those from minority backgrounds for AI policy roles. Mentorships, especially informal ones, empower women in AI with guidance to maximize opportunities and overcome obstacles.
Build a robust pipeline to make this happen while keeping coordination and flexibility as the key. Businesses should educate women on the diverse roles and continuous learning opportunities in AI beyond technical fields, and recruit those with aptitude from a variety of backgrounds. Removing ambiguity about pathways into AI can attract overlooked female talent and skill sets needed for responsible innovation. Also, promoting female AI role models and leaders can inspire young women to pursue STEM, while showcasing an organization's commitment to gender inclusion. Visible examples of women's success in AI and its governance help shift cultural stereotypes and attract future contributors. And then sponsoring programs to get more girls and young women interested in STEM fields relevant to AI, through education initiatives and work experience can make a difference. In addition to this, removing biased barriers to advancement like lack of flexible work options or parental leave that disproportionately affect women will lead to family-friendly policies and will seem inviting for women.
There are many measures which can be taken by governments to bring women’s perspectives into policies. Funding research by diverse scholars on AI ethics, law, and policy to expand the pool of women experts and listening to the perspectives of women's organizations, minorities, and civil society groups in the policymaking process. Also, the promotion of inclusive cultures within organizations that develop AI and welcoming diverse leadership. Major step is to publicly commit to diversity, equity and inclusion in AI policymaking at the highest levels of government and industry and then increasing transparency about current representation and setting goals to measure progress over time.
There is an emerging field in AI governance called ‘Feminist AI governance’ which combines policy, development and research to maintain the standards of equality and inclusivity during the AI life cycle.
This field will evaluate the imbalances and inequalities prevailing in the AI ecosystem, and harms perpetuated by this to address these issues globally. Initiatives are being undertaken in this regard across various countries. Supporting these will help a community of feminist scholars in AI governance to emerge, which will eventually shape an inclusive and strong tech-oriented society for the future.
Conclusion
The lack of female perspectives and participation in artificial intelligence governance puts society at risk of biased, unethical systems. Women face multiple barriers in accessing STEM education and careers necessary to shape AI policy. Without deliberate efforts to recruit, retain, and promote women in AI, diversity will continue declining. Organizations and governments must take tangible steps to value women's contributions, eliminate discrimination, and prioritize inclusion across the AI field. Achieving equitable gender representation in AI development and policy is crucial to harness AI's benefits for all. The perspectives and insights women provide are indispensable in steering these influential technologies towards a just future. But this requires fundamentally rethinking how women are treated and valued. True change begins with meaningful female participation at all levels of AI research, development, and governance.
Comments