19 Dec 2018PopsciHow Google tackles bias by dropping gendered translationsPOPSCI: A scientific slant on popular culture
image-d5ee60fbbf2fdb0a0ae7717e626dfee6d7f6c3b4-1349x470-jpg

Google is working towards removing gender bias from Google Translate, by providing both feminine and masculine translations instead of just the masculine. Bias in tech is something that is proving to be an obstacle to wider uptake, especially if it's insensitive to evolving gender norms. We explore the insights behind how dropping masculine as the default is tackling gender norms and biases in our everyday communication.

Author
Sophie Robinson Sophie Robinson is a behavioural analyst for Canvas8. With a background in anthropology, she’s experienced in understanding the cultural mechanisms that shape the world. When not working, she’s making documentaries for her MA programme or wild swimming.

Traditionally, Google's translation tool has been biased towards 'masculine translation', but Google has said it will be providing feminine translations, regardless of gendered connotations, where possible. The removed gender bias tool is available when translating English into French, Italian, Portuguese, and Spanish and Google has plans to extend its gender-inclusive translations across more languages and for iOS and Android platforms. Google Translate learns its translations from hundreds of millions of already translated words on the internet, but the tech brand wants to take it further by utilising AI to help the system "addresses non-binary gender in translations," so says a Google blog post.

A biased translation could reinforce negative stereotypesStrelka Institute for Media (2017)

Google is recognizing that for many people, gender is a social construct – and there is a growing expectation for the tech services they use not to be gendered by default, or maintain damaging human biases. And with 90% of US parents believing communication is the most valuable skill for children, the removal of gender bias aligns the tech brand with the progressive side of the gender movement without moralising or proselytizing. Initiatives that show the brand's softer touch are reflected in its introduction of emotional intelligence to its AI systems, teaching machines to be familiar with emotional expressions.

Sophie Robinson is a research intern and recent social anthropology graduate at Canvas8, which specialises in behavioural insights and consumer research.