Prediction vs. Explanation and its Effect on Statistical Methods [duplicate]

Question

In layman's terms, what is the difference between predicting and explaining in statistics? I was looking for the differences between AIC and BIC and found this post with an answer stating:

My quick explanation is

AIC is best for prediction as it is asymptotically equivalent to cross-validation. BIC is best for explanation as it is allows consistent estimation of the underlying data generating process.

This makes me think that in the same vein as precision and accuracy there is some core distinction here that significantly effects when and how to use a lot of statistical procedures. I googled it and only found a variety of papers including this one, but this is far too rigorous for my question given my current knowledge. Could anyone provide an intuitive exposition of the difference and perhaps some examples of how it affects the use of statistical methods and tools?

Answer 1

If you care about prediction you don't care why something happens you just want a cristal ball that will warn you if event x is likely to happen for subject y. If you care about explaining than your priority is to find the "story" of why things happen; accurate prediction of the future is of secondary interest.

In a true model there would be no contradiction between prediction and explanation. However, it is logically impossible for a model to be true: the definition of a model is "a simplification of reality" and a simplificaiton is "wrong in some useful way", so an object cannot be simultaneously a model and true. What is "useful" in "wrong in some useful way" is different depending on whether your primary interest is in prediction or explanation.