Science has progressed by reasoning on what models could not predict because they were missing important ingredients. And yet without correct models, standard statistical methods for scientific evidence are not sound. Here, I argue that machine-learning methodology provides solutions to ground reasoning about empirically evidence more on models' predictions, and less on their ingredients. Science uses false models as means for truer theory [Wimsatt, 1987]. How can statistical tools ground valid reasoning on empirical evidence without true models? Generalization is the key. Here I develop the argument that, unlike popular belief, reasoning from black-box models is good for science, because it builds on the validity of inferences on prediction of observables.