Reading through the replies it's clear half are worse than GPT-4's performance on the prompt