They don't demonstrate anything remotely similar to knowledge. They demonstrate the fundamentally nonlocal, "vague" properties of the fundamental constituents of reality. After all, we can use experiments like the delayed-choice experiment paradigm to demonstrate that we can determine the results of experiments depending upon the ways we wish to determine them (i.e., if we want a system to behave more like a classical particle, we can do this, and likewise for a classical wave; in both cases, WE determine what we observe and there is no indication whatsoever that what we observe "knows" anything at all, but much to indicate that this is impossible and nonsensical).