1 paper
Timothy Kassis, Vinayak Agarwal, Yuhuan He +2
A language-model agent asked to analyse an experiment will usually return working code. Whether the analysis is defensible is a different question. A defensible analysis depends on…