Bibliography (4):

  1. GPT-4 Technical Report § Harms of representation, allocation, and quality of service [loss of humor]

  2. https://openai.com/index/gpt-4-research/

  3. Deep reinforcement learning from human preferences