So, while cool they have generally been replaced by far more useful and less complex topics.
One example is if your looking for duplicates among huge data sets. You farm out your Filter to 1,000 machines merge them, then re run to get a list of possible collisions. Then compare that list. But, well merge sort can do the same thing a little slower, it's a rare problem, and the pathological case for Bloom filters is really bad. So, even if you know about Bloom filters you may go with the simpler idea and be done sooner.
PS: In that example hash speed might mean sorting is actually faster.