It will, being a deterministic machine, generate any kind of wrongthink that is in its training data. Ironically, all of the media coverage of negative stereotypes by well intentioned activists probably even makes it more likely to generate this kind of data.