Fairness and languages left out
Stories
AI and training data · Brief
AI and training dataAI and training data · Brief
AI and training dataAI and training data · Brief
AI and training dataAI and training data · Brief
AI and training dataAI and training data · Brief
AI and training dataAI and training data · Brief
When three readers disagree, the label hides it
A new Arabic dataset keeps every reader's verdict, and shows what a single majority label throws away.
AI and training dataFourteen experts label bias in 448 Hindi news articles
A small open data set shows how a language left out of the tools gets its first measuring stick.
AI and training dataThe most accurate post moderator is often not the fairest
A new test set shows that judging posts well and judging them fairly are different skills.
AI and training dataIgala gets open translation data that records dialect
A small sentence set for a Nigerian language shows why the mix of dialects in data changes results.
AI and training dataReweighting training data can shift harm to another group
A preprint finds a common fairness fix lowers missed cases for one group by raising them for another.
AI and training dataA new corpus tests abuse detection in three languages
Detectors built on Northern data miss online abuse of women in Levantine Arabic, African French and Swahili.