With the advancement of large language models, I am curious about their implications for the industry practice of classification tasks, where traditional models like logistic regression and tree-based models currently dominate.