ML is useful for many things, but not for predicting scientific replicability

L: https://www.aisnakeoil.com/p/machine-learning-is-useful-for-many

C: https://news.ycombinator.com/item?id=37089573