Nature Podcast

How stereotypes shape AI – and what that means for the future of hiring

Informações:

Sinopse

00:48 The stereotypes hidden in Internet imagesStereotyped assumptions about women’s ages and their perceived job suitability are enhanced by Internet imagery, according to new research. A study of hundreds of thousands of online images shows that women appear younger than men. This stereotype extends to the jobs that people perceive women do, with men being associated with roles such as CEO or head of research, while women were linked to occupations like cook or nurse. The research shows that these biases have been embedded into the training data for AI models and could affect future hiring. The researchers caution that society is at risk of creating a ‘self-fulfilling prophecy’ where these stereotypes shape the real world.Research Article: Guilbeault et al.News and Views: Distorted representations of age and gender are reflected in AI models13:24 Research HighlightsA very hungry planet — plus, how climate change is leading to larger trees in the Amazon.Research Highlight: ‘Rogue’ planet is fa