P=NP is referring to two sets: P is the set of problems which can be solved with a polynomial time algorithm, and NP is the set of problems which can be checked to see if the solution is correct in polynomial time, but a solution can't be found in polynomial time.
As an example for how a distinction is natural:
For example, if I asked you to find integer solutions to the equation xy + yx=145, this would be fairly difficult. But if I tell you x=3, y=4 is a solution, it's really easy to check.
It's a famous conjecture that P is NOT equal to NP: in normal language, that just because a problem is easy to check, it doesn't mean it's easy to solve. Nobody actually has a proof one way or the other though