I think there are two different sorts of “prescriptivism” that need to be distinguished.

One is based on the idea that every language has one ideal form that must be elucidated by experts and enforced across the rest of the language community. This was never a real part of linguistics—it’s a popular misconception from people imagining that linguistics is an extrapolation of grade-school grammar.

But the other form of “prescriptivism” is just fluent speakers correcting people acquiring the language when they perceive a mistake—this is a natural function that occurs in all language communities, starting with parents teaching their children. And it’s a necessary function: learning is orders of magnitude more efficient when the learner has both positive and negative examples.

The problem comes when speakers miscategorize speech in other dialects as incorrect examples of their own dialect, and attempt to correct them accordingly.