[2021]
BABY is a language model trained on scraped data from social media influencers, over 50,000 posts from Instagram, Facebook, and blogs, combined with poetry, lyrics, and haiku databases. Fine-tuned to isolate the tone, content, and structure of influencer speech and recompose it as generated text. Removing personal authorship, BABY makes the structure of influence and performativity visible. Deployed as a standalone desktop application.
Conceived and led by Rhea Dally. Developed with Guillaume De Sa, Marco Zausch, and Tim Seifert during the Le Wagon Data Science program (scholarship, 2021).
Related Work