Companies competing in the chatbot wars are using something known in the industry as “the Pile” to train their large language models. It’s a trove of open-source data made up of text scraped from all around the internet, including Wikipedia and the European Parliament. Annie Gilbertson, investigative reporter for Proof News, recently took a deep dive into the Pile and discovered something else: a dataset called “YouTube Subtitles.” Marketplace’s Lily Jamali spoke with Gilbertson about her investigation and how YouTube creators feel about their content being used without their consent.
