Genevera discusses the complexities of conditional inference in machine learning, emphasizing the analytical challenges involved in decomposing test statistics. She highlights a disconnect between machine learning researchers and practitioners, noting the need for clearer communication and improved techniques that cater to scientists using these methods for data analysis.
