I presume you mean that m is usually positively correlated with p over some unspecified time horizon?
A problem I often find with economic truisms is that they tend to carefully ignore the time component. Over a short enough time horizon this is clearly untrue. Over a long enough one it's pretty much tautological. Market theory has a similar property.
> if you print too much money, you will get inflation
Again, usually. Hasn't really been happening over the last few years in the US and Europe. Presumably we just wait for some arbitrary long period until inflation starts to rise again? Does the theory tell us how long?
So it's not that economics can't make predictions. It's just that they're often not useful predictions. It is quite good at postdictions, of course. The utility of which is at least a tad suspect.