PEOPLE ARE JUST AS BAD AS MY LLMS

Summary from the TLDR Web Dev newsletter

Both large language models and humans have biases in pairwise comparison ranking tasks, with LLMs showing preference for the first item presented and humans showing preference for the item on the right.