I think what you and the writer of the article are complaining about is not the LLM but the dataset that the LLM is describing. My guess is that if you created

I think what you and the writer of the article are complaining about is not the LLM but the dataset that the LLM is describing. My guess is that if you created a GPT of a well-regarded book on WW1, you’d be able to have an enlightening conversation about the topic.