Thus, we think that the reproducibility crisis in machine-learning-based science is still in its early days. With some studies now using large language models in research — for instance ...