As Artificial Intelligence expands into all areas of life, UN Women has made an important call for wider adoption of gender-sensitive AI. The appeal goes beyond general caution; it stems from a study revealing that algorithmic bias is reinforcing systemic gender inequalities, leaving women increasingly vulnerable to discrimination and violence. The problem has been built into the technology. When systems learn from historical, flawed data, they tend to replicate and amplify societal prejudices. An analysis of 133 generative AI systems found that 44% demonstrated clear gender bias, while 26% exhibited a combination of both gender and racial bias.

The study found that AI actively absorbs human prejudices, racial biases, and tendencies to discriminate based on gender. These biases are not mere algorithmic glitches; they are systemic patterns reflecting deep-seated societal flaws. Large Language models (LLMs) routinely perpetuate stereotypes, linking women with home, family, and children, and men with careers, business, executive roles, and salaries. Many systems have also been seen as exhibiting sexist or misogynistic attitudes, reducing women to sex objects or the property of their husbands. UN Women noted that these toxic outputs are the natural outcome of training AI on decades of unequal data. Global policy has failed to keep pace with these emerging realities. Out of the 138 countries assessed in the analysis, only 24 referenced gender in their national AI strategies, and only 18 adopted substantive gender-responsive measures. UN Women sees this as a choice made “over and over” in training data, design rooms, and policy documents “that stay silent on half of the population”.

Under-representation of women across the AI ecosystem, particularly in decision-making roles managing core technology and product development, has been cited as a key reason for AI systems inheriting gender bias. UN Women notes that women risk bearing higher costs of AI adoption, as they face a bigger threat of losing jobs to automation and are more vulnerable to AI-enabled abuse. Addressing these inequities requires systematic gender- and race-based audits of AI models and the output they generate. Investment must be enhanced towards equipping women with AI skills; their representation in leadership roles is critical in shaping gender-sensitive technological decisions. UN Women has urged governments, companies, and experts researching and developing AI to integrate gender equality throughout the AI life cycle, from design and development to deployment and governance. The group has rightly noted that “when designed with safety and used with intention,” AI can help detect stereotypes, broaden representation, and improve accessibility at scale. (Source: DH)