Understanding the 'knowledge' and biases within LLMs through their pre-trained data, and considering ethical implications for society.