[tarruda]

LLMs are not magic, they inherit whatever bias is in its training data.

[xvector]

Their training data is clearly better than whatever most journalists train on.