
"AI systems trained on historically biased data could replicate and even amplify existing discrimination in areas such as recruitment, career progression and performance evaluation. Without more representative datasets, stronger oversight and greater diversity among the people designing and deploying AI systems, the technology risks embedding workplace inequalities at scale."
"One case involved the withdrawal of an AI recruitment tool developed by Amazon after it was found to favour male candidates over female applicants. Concerns have also been raised about the visibility of women's professional content on platforms such as LinkedIn, where algorithmic ranking has reportedly reduced the reach of posts written by women."
"Women face a dual risk from the rapid expansion of artificial intelligence: they are underrepresented in the development and leadership of the technology sector, yet are overrepresented in roles most vulnerable to automation. Administrative, education, healthcare and social care positions, many of which are dominated by female workers, face significant automation risk."
Research from the Women and Work All-Party Parliamentary Group reveals that artificial intelligence could deepen gender inequality in workplaces without increased female participation in technology design. AI systems trained on historically biased data may replicate and amplify discrimination in recruitment, career progression, and performance evaluation. Real-world examples demonstrate algorithmic bias, including Amazon's recruitment tool favoring male candidates and LinkedIn's algorithms reducing visibility of women's professional content. Large language models learn patterns from historical employment data reflecting gender imbalances, potentially reinforcing these inequalities during deployment. Women face dual risk: underrepresentation in AI development and leadership while being overrepresented in roles most vulnerable to automation, including administrative, education, healthcare, and social care positions.
#ai-bias-and-discrimination #gender-inequality-in-technology #algorithmic-decision-making #workplace-automation #ai-development-diversity
Read at Business Matters
Unable to calculate read time
Collection
[
|
...
]