Because despite not realizing it yet, human consciousness is just a next-word-predicting LLM, and the dataset most Americans are trained on has those as the most common next words.