@SpirosMargaris
A startup is paying people to “bully” AI chatbots. The job is to push systems until they break, exposing where they forget, hallucinate or contradict themselves. It sounds unusual, but it highlights a serious point: stress-testing AI is becoming a real profession. If AI is going to be everywhere, who is responsible for finding its weaknesses before it matters? https://t.co/7zmclNEIBQ