Yes, they are important. Both mocks and fakes (apologies for the terminology; I don't know if there are widely-used terms that distinguish) stand in for real collaborators, but mocking frameworks can't really provide arbitrary functionality and so lead the developer to write unit tests that do what the mocks can do: simply verify methods called, produce specific returns, etc. This tends to lead to unit test code that verifies the wrong things and is very brittle. Fakes, or ad-hoc mocks, encourage the developer to check what needs to be checked and behave in the minimal necessary ways.
In Java, creating fakes is often hard and sometimes impossible. In Python, it's always possible and usually easy. I hope unittest.mock doesn't lead developers to feel that it is the right way to mock out their collaborators, rather than using more flexible and precisely-tailored ad-hoc mocks (which I call fakes).