Once you accept that a model of reality is always somewhat incorrect you kind of stop seeing this as "lies". Reality is infinitely complex, a model however is finite and therefore understandable.

We value models for their simplicity, for their ability to disregard secondary and tertiary details to allow comprehension and prediction of primary effects.

Then the only question is: which level of the simplicity/correctness trade-off is appropriate in given circumstances? When teaching kids or even a non-professional adults, the appropriate level is often very heavy on the "simplicity" side. Adding more complexity results in less overall understanding, so a net-negative effect in the end.