Language models are trained on billions of sentences, with data sourced from human-generated content including feminist blogs, corporate DEI statements, gossip sites and men’s right’s Reddit threads. So, what does this mean for how gender is handled by AI? www.ox.ac.uk/news/feature...

The model was trained “using large volumes of telephone conversations and text messages, obtained through its [Unit 8200’s] extensive surveillance of the occupied territories.”

New blog! @oiioxford.bsky.social doctoral researchers @deeliu97.bsky.social, @manueltonneau.bsky.social and Juliette Zaccour propose a series of recommendations for effective data access and data governance in light of the EU’s Digital Service Act. Read the full article: bit.ly/49DJyhn

Algorithms are now making critical decisions across sectors. And yet, independent investigations of discriminatory decisions are stymied by strong barriers to data access. Our new research article just accepted at CHI establishes (minimum) data access requirements for rigorous audits.