There is an epistemic difference between knowledge of the past/present and knowledge of the future. In the cases where we can predict the future with a high confidence level, I still don't think this would count as knowledge, mainly for the reason that any proposition which involves a prediction where there is a chance that it turns out to be false cannot be regarded as true (at the time it is made). think I agree with the basic JTB formulation, i.e. the proposition has to be true and justified at the time it is made.
I accept that some knowledge that we previously held to be true can subsequently turn out to be false, and I don't think that prevents the old knowledge from being classified as knowledge at the time, or means that we cannot in fact know anything.
I'm not inclined to adopt the approach of only really knowing things with a high degree of probability. For example, I don't think that it is only highly likely that the Earth orbits the Sun, I think this is 100% true. If, for some reason an alternative theory was proposed which negated the heliocentric view, then I would change my assessment at the time.