Sometimes we don't know the improvements until we have done enough experiments to provide a model that aids in discovering improvements. Electricity was hard (even fatal) to experiment with before it was understood (and frontier work in electricity when running at extreme frequencies or energies is still difficult to understand, and if we're counting all EM in there it goes beyond being easy to manipulate and experiment with).

My guess? If we figure out how to detect dark matter we can get closer to figuring out how to interact with it (other than through the very very very weak gravitational force). Or maybe we figure out that it was a spinning universal frame or something that gives us a better Standard Model.

If, however, we figure out how to interact with it and can harness any potential energy from it, then by definition we won't see any interference in the electromagnetic forces. That would be incredible, that would be as good as having readily available superconductors.

To wit, consider the case of Charles Parsons, inventor of the steam turbine engine. He was able to do this because of the lifetime work of Henri Victor Regnault (https://en.wikipedia.org/wiki/Henri_Victor_Regnault) characterizing the behavior of steam under pretty much every possible regime. At the time Regnault's work was largely ignored, and later forgotten, since it was only of minor academic interest.

It was only years after his death that Parsons would use his work to determine the necessary geometry of the rotors and stators in his new and revolutionary engine. Regnault certainly didn't see that coming, the people who overlooked his work didn't either, but that pure science for the sake of science helped to change the world.

So as you say, we often don't realize what we're going to do with the results of pure science until engineering catches up, sometimes decades or centuries later. Still without the pure science we'd never get the engineering.