ai researcher uncover
AI researchers uncover ethical, legal risks to using popular data sets
Hugging Face has found that data sets have better documentation when they are open, consistently used, and shared, said Yacine Jernite, leader of its machine learning and society team. The open source company has prioritized efforts, like automatically suggesting meta data, to improve documentation. Even with imperfect annotation, openly accessible data sets are the first meaningful step toward more transparency in the field, he said.