Playing Chess Against LLMs And The Mystery Of Instruct Models

At first glance, trying to play chess against a large language model (LLM) seems like a daft idea, as its weighted nodes have, at most, been trained on some chess-adjacent texts. It has no concept …