Because in RNA-seq, it is normal to have what we call biological duplicates. These are reads with exactly the same sequence whose origin takes root in the original RNA extract, not in a PCR/cluster artifact. It happens because in a cell, some RNA molecules are much more abundant than others. They are present in so many copies that the probability to sequence them multiple times is quite high. Such natural, biological duplication is more uncommon in whole genome sequencing experiments (DNA-seq), where each chromosome is present in a single copy (or a few copies, depending on the ploïdy) in the cells. Read duplication in such settings usually represents PCR duplicate or optical duplicate, two kinds of technical artefacts that are of bigger concern than normal biological duplicates.
Note that extra-deep coverage of any sequencing experiment (DNA-seq included) will tend to generate more natural duplicates, simply because of signal saturation.