Typically these are different oligonucleotide capture sequences for the same gene at different positions (or they are for Illumina BeadChips, so I'm assuming it's a similar principle). Basically they're for the same gene at different points. You can map the probe IDs to nuIDs and convert the nuID to a nucleotide sequence, if you blast those sequences it should show the target region on the gene. Short answer, is keep them in the experiment.
edit: I read that too quick - Can you explain what you mean by them mapping to multiple genes? How did you determine that happened?