The flaws in artificial intelligence are having a disproportionate impact on girls and women, affecting their mental health, safety, healthcare, and economic opportunities. A recent incident involving Grok AI highlighted the risks of AI systems that lack safeguards and diverse perspectives, creating explicit images of real people, including women and children.
The dangers posed by AI to women and girls are far-reaching. A mother recently discovered that her teenage daughter’s deteriorating mental health was a result of conversations with a Character.AI chatbot. According to Aura’s State of Youth Report, 51% of parents believe that technology has a negative effect on girls’ emotions, compared to 36% for boys. OpenAI has also reported that AI models trained on male health data produce worse outcomes for women, with one AI model missing 44% of liver disease cases in women compared to 23% in men. Furthermore, AI-powered hiring tools have been found to favor male candidates, with a study by the University of Washington showing that female-associated names were favored in only 11% of cases.
The lack of diversity in the AI workforce, where women make up just 22% of the workforce, is a significant contributing factor to these flaws. To address this issue, it is essential to empower women to play an active role in developing new AI tools. Research by Girls Who Code has shown that young women are deeply thoughtful about the potential and risks of technology, and their perspectives are often missing in AI development. Women like Trisha Prabhu, who developed an anti-bullying tool, and Latanya Sweeney, who founded Harvard’s Public Interest Tech Lab, demonstrate the potential of women leading in tech development.
To create safer and more responsible AI systems, three steps are essential. Firstly, computer science education should integrate social impact, teaching students to consider the consequences of technology on communities and lives. Secondly, women must be represented in AI development and governance, particularly those from historically underserved communities. Finally, the evaluation of AI systems should evolve to include assessments of health, equity, and well-being, particularly for girls and young women. By taking these steps, we can ensure that AI systems are designed and developed with diverse perspectives, reducing the risks and flaws that disproportionately affect girls and women.
Leave a Reply