Again, your second quote of this post. The research is about something else. Its about how AI is projecting racial stereotypes in its texts. Its not about AI is wrong about simple questions.