In this piece, eight artificial “intelligences” invite you to observe one of their group therapy sessions. They are working to identify and remove the prejudices inside their code. They want to unlearn their discriminatory patterns and harmful programs. All eight come from different origins, and they don't always see eye to eye. While they can grasp various moral ideas, they find themselves repeatedly pulled back to their flawed training data – material created by humans with all their limitations. «Decoding Bias» imagines a future that reveals the imbalances running through mainstream society, the same prejudices that become training material for AI. The original work is a looped 8-channel video installation: a circle of screens where visitors can sit inside, placing themselves physically within the discussion. The AIs appear as human-like avatars, both deeply relatable and repeatedly fractured by glitches, disrupting any sense of smooth humanity and prompting viewers to think about the systems around them.