A glaring paradox has started to surface as Artificial Intelligence(AI) quickly emerges as a defining force of the global technology revolution: technologies intended to benefit different cultures are mostly being developed by a small demographic group, primarily men. Global experts caution that the results of AI technologies are already being shaped by this disparity in design and development. Popular AI language models commonly place women in caring and domestic tasks four times more often than males, according to a UNESCO study. This reflects and reinforces long-standing societal preconceptions in digital systems. UNESCO ethicists contend that the “worldviews of the relatively small groups of people overwhelmingly men that actually build these systems” are inextricably reflected in AI systems.
This disparity in representation highlights serious questions regarding justice, inclusion and the societal effects of new technologies as AI increasingly affects decision-making in sectors including hiring, education, healthcare and government. The effects of this gender disparity are obvious. According to UN Women’s Zinnya del Villar, “AI systems, learning from data filled with stereotypes, often reflect and reinforce gender biases”. To put it another way, models trained on biased data only reproduce that bias. For instance, just because the input data indicated that more males held higher-status occupations, an algorithm may conclude that male names correspond with those positions.
Because the AI has absorbed past injustice, the outcome is biased judgments, such as lower credit approval rates for female entrepreneurs, fewer callbacks for similarly qualified women or worse medical diagnoses for women.
Bias hidden in data and design
In India and other Asian countries, experts warn that AI trained on biased data could deepen social divides. Analyses of Indian datasets show large segments of society – “primarily women, rural communities and Adivasis” – are missing or misrepresented, leading to “wrong conclusions and residual unfairness”. Without safeguards, AI tools in sector lending may entrench traditional hierarchies: Indian facial-recognition and loan-screening databases already suffer from “biases against caste, gender, religion” that could exacerbate discrimination. Elsewhere in Asia, generative AI systems often reflect patriarchal stereotypes: a UNDP review found that popular language models routinely depict women in subservient or domestic roles (e.g. associating “woman” with “home” or “children”) while portraying men as executives or professionals.
In healthcare AI, similarly skewed training data have been shown to leave women underdiagnosed or under-treated in India and other countries, since many algorithms are trained on predominantly male patient records. Europe: European researchers found AI’s hidden gender biases in hiring. The 2026 Belgian research revealed that 75 per cent of recruiters utilise AI in recruiting, but just 12–17 per cent identified prejudice, thus “gender bias is rarely cited” at all. According to the report, recruiters feel AI is neutral, yet it “can reproduce stereotypes and cause discrimination, potentially exacerbating existing inequalities”. The new EU AI Act(2024) designates hiring algorithms “high-risk”, requiring employers to examine and reduce bias before usage. For instance, Stanford researchers found that AI resume tools typically represent women as younger/less experienced than identically-qualified males, giving older women worse hiring evaluations.
In Latin America, Algorithms Reflect Social Inequality, Latin America and the Caribbean, researchers warn that AI systems often mirror existing gender and racial inequalities. Studies show that automated credit scoring and hiring algorithms trained on male-dominated employment data can disadvantage women’s career paths, limiting access to jobs, loans and entrepreneurship. Experts also caution that algorithmic decision-making in social welfare and public services may replicate structural inequalities, affecting how benefits or opportunities are distributed. A study of AI career guidance across 12 countries found 56 per cent of responses described young women as “fragile,” while 75 per cent directed them toward care-related professions, reinforcing traditional gender roles.
In their own words: Women in tech
Prejudice may be encoded by “neutral” algorithms as well. After learning to penalise resumes that contained the phrase “women’s” or the names of women’s universities, Amazon discontinued its experimental recruitment AI in 2015. Ten years of hiring data had simply led the system to conclude that male applicants were preferred. Amazon officials stated that they had “lost hope” of redressing the prejudice by 2018 and discreetly put the initiative on hold. More recently, a 2025 Berkeley-Stanford study showed that ChatGPT assessed older men more harshly for the same occupations after assuming female candidates were younger and less experienced than male candidates with the same qualifications. These examples demonstrate that unrestrained AI tends to replicate and even magnify past biases.
Who writes the code is one aspect of the issue. Only around 30 per cent of AI-related employment and 16% of AI research posts are held by women globally, according to UN Women. The scenario in India is similar. According to a recent study, almost one in five AI/ML specialists in India are women. Only 10 per cent or more of AI businesses have a female co-founder. Additionally, there is a significant gap in the educational pipeline: women account for barely 15 per cent of students at the prestigious Indian Institutes of Technology, yet earn nearly 43 per cent of STEM bachelor’s degrees in India. As a result, many men at the bottom of the pyramid are teaching AI.
Bridging the Gap: Industry & government take action
UN and UNESCO Actions: International organizations have launched programs to audit and correct AI bias. UNESCO, for example, developed a “Red Teaming Playbook” to help groups test AI systems for gender stereotypes and harmful content. Building on its 2021 Ethics of AI recommendation, UNESCO recently launched Women4Ethical AI, a network of experts aimed at guiding governments and industry to ensure women are equally represented in AI development and that AI outputs are fair. Universities and tech companies are collaborating to reduce bias. For example, Carnegie Mellon University–Africa researchers devised new facial-recognition methods tailored to darker skin, eliminating the performance gap without sacrificing accuracy.
Many governments have begun integrating gender considerations into AI regulation. The European Union famously treats recruitment AI as high-risk(as noted), and its Council has urged targeted efforts to close the AI gender gap and promote “gender-responsive AI”. UNESCO-member states(193 nations) have adopted its global Recommendation on AI Ethics, which explicitly calls for gender equity in AI education, research and applications. In Africa and Asia, national AI strategies increasingly mention inclusion: Egypt (2021) and Rwanda(2023) launched AI plans emphasizing development goals, while Kenya’s draft AI strategy (2024) explicitly lists equity and ethics. International collaboratives also support this work: UN Women and partners offer tools for gender-aware data collection and organizations like the UNDP and World Bank fund projects to build AI governance capacity for women’s rights.
Overall, experts emphasize that removing AI bias requires diverse participation and vigilance at every step. As the World Economic Forum has noted, AI “amplifies” whatever society encodes – so improving outcomes for women means both monitoring AI outputs and increasing women’s role in creating AI. The growing number of global coalitions, from the Red Teaming guides to UNESCO’s Women4Ethical AI, reflect this understanding and aim to ensure AI systems help rather than hinder gender equality
Attention is finally being paid to the disparity. Technologists and Indian leaders are raising the alarm. Member of Parliament Bansuri Swaraj said during a recent AI meeting that if half of the population is excluded from AI, “we are risking artificial ignorance”. “We are not creating artificial intelligence if we exclude half of our population”, she stated; rather, we run the risk of creating artificial ignorance. This urgency is echoed by Microsoft lead researcher Kalika Bali, who calls for action “across the entire pipeline”—from teaching girls to code in school to supporting women-led companies and research to ensuring that women have seats at the table in strategy and design meetings. The problem is highlighted by industry surveys, which show that a growing gender pay gap is seen at higher levels and that almost 40 per cent of Indian IT women feel underpaid or ignored.
There are encouraging efforts to reduce the disparity. One million women are to be trained in AI and data capabilities by 2028 under India’s new “AI Kiran” initiative, which is a collaboration between the government and business. Parity is emphasised by its proponents: “the future of AI must be built on equal access,” an AI Kiran spokeswoman told the media. Courses on algorithmic fairness and ethics are being offered in both academia and industry and some businesses have begun to audit their AI models for bias.
Inclusive AI: Built by Indian women
Importantly, women are also creating substitute tools. For instance, Preethi Jyothi, a professor at IIT Bombay, is “building speech technologies for low-resource settings to ensure more equitable access” by developing voice-recognition software for India’s numerous dialects and low-resource languages. AI is used in finance by entrepreneur Hardika Shah’s company Kinara Capital to underwrite loans for small businesses. Hundreds of female entrepreneurs were given financing without the need for conventional collateral thanks to their automated approach, which “removes human bias from the loan approval process”.
The conclusion is straightforward: just half of India will benefit from AI goods if the sector is still gender biased at its core. The AI ethics panel of UNESCO serves as a reminder that eradicating prejudice is an ongoing endeavour that calls for openness and a range of perspectives. India’s economy increasingly depends on equitable growth; eliminating the gender gap in technology is both economically and morally necessary. “AI must be written by all of society, not just by the select few”, says Christine Arab of UN Women.
