For human beings, we have two copies of each gene inherited from the parents. The question is, when referring to gene expression, which copy (or allele) is considered?
Genome assemblies usually feature a haploid set of genes, unless particular effort is applied to assemble separately the homeologs. With that said, the gene copy you have in the assembly is usually one of the two, and usually the one you had more genomic reads for.
In terms of gene expression, if you measure it by RNASeq you will get reads from both and both of the read pools will map on the same gene. The ones that come directly from it will map with no mismatches (ideally) and the ones that come from the homeolog chromosome will map with a few mismatches (probably).
The expression of the "gene" will therefore be the expression of both genes, summed together. If we consider it as the same gene, though, it doesn't really make a difference if you sum them up or count them twice separately; unless we are talking about specific homeolog silencing like the Xist mechanism, in which it is important to remember that one copy of the two is off.