Twitter find its artificial intelligence tools tend to cut out black people from photos


The company said in a new study Wednesday that Twitter’s image cropping algorithm has a problematic bias in excluding blacks and men, adding: “How to crop an image is people’s best decision.” Three in the studio. 

A machine learning researcher who criticized the image preview in a post last year that users criticized for not including black faces and found that, compared to demographic equality, there is an 8% difference in the preferred rate of women and a difference of 4% in the white population. 

Reasons include issues with an image background and eye color, but none of them is an excuse. “Machine learning-based clipping is fundamentally flawed, removing user agents and restricting users’ expression of their identities and values, rather than enforcing rules. 

The researchers wrote, noting which part of the image is more interesting. The researchers also evaluated whether the crops favor women’s physical expenditure, reflecting the so-called “male eyes,” but this does not appear to be the case. The researchers also evaluated whether the crops are beneficial for the women’s physical expenditure. 

The document says: This finding is another example of the different effects of artificial intelligence systems, including the demographic biases identified in facial recognition and text analysis. 

Researchers from Microsoft and MIT in 2018 and a subsequent study by the government of In the US, facial analysis systems were found to be easier to identify people of color than people of color. In 2018, Amazon canceled AI recruiting tools that were biased against women.

Leave a Comment

Your email address will not be published.