Bias can be learned and perpetuated in different ways (i.e., societal beliefs, misrepresentation, ignorance) that consequently create inequitable outcomes across all spheres of life.
TakeTwo aims to help mitigate bias in digital content, whether it is overt or subtle, with a focus on text across news articles, headlines, web pages, blogs, and even code. TakeTwo will leverage a crowd-sourced database of words and phrases that could be viewed as racially biased in the US.
Verified and trusted contributors can use TakeTwo browser extension to select potentially biased language in text-based media. These selections are classified under commonly detected types of racially biased language to train TakeTwo text-classification machine learning model.