Open Studio: Can Bias in LLMs Be Used for Good? | Francesca Lucchini Wortzman
Featured
It is a well-known fact that LLMs express harmful biases in their predictions. The main source comes from the training datasets, which are too large and expensive to check thoroughly. In this open studio, Francesca Lucchini will explore how we can leverage the bias in LLMs and use them to examine massive datasets, discovering starting points for a data audit.