Patchwork

Mimi and Chee discuss the design choices behind large language models, arguing that LLMs do not need to generate human-like language and that their personalities are a training choice, not a necessity…