Can we trust scientific discoveries made using machine learning?

Statisticians are cautioning fellow scientists not to make assumptions about the accuracy, uncertainty or reproducibility of scientific discoveries made with today’s machine learning models.