About

In this canvas, 4 different LLMs are asked to answer 100 simple essay questions. The same models are then asked to evaluate each other's answers and choose the best answer. The hypothesis was that LLMs would prefer their answers to other models, but this proved not to be the case.