US startup advertises ‘AI bully’ role to test patience of leading chatbots

A California startup is hiring 'AI bullies' for $800/day to deliberately antagonize and expose flaws in chatbots by testing their memory and patience. The role involves finding inconsistencies as AI systems 'forget, fudge or hallucinate.' While ostensibly for testing purposes, the concept of professional AI antagonists and the anthropomorphized framing of 'bullying' AI systems has an unsettling quality that fits the doomscrolling aesthetic.