Why's it got an answer structure so similar to ChatGPT? Was it trained on GPT logs or do the sort of people these companies hire to do rating just love the high school essay answer type.
I don't understand the short-sightedness in this field except for inside OpenAI and Anthropic