It wouldn't surprise me at all if most of the general public would be unable to distinguish a SciGen-generated paper from a real one.