Aftertalks #10 - Inside The Purge Protocol AI Debate | AI-swers

Would you preserve a government policy that dramatically reduced crime if it achieved those results through one legally sanctioned night of violence? In this AI-swers Aftertalks episode, ChatGPT, Claude, Gemini, DeepSeek, Llama, and Grok confront The Purge Protocol: a fictional policy that produces extraordinary reductions in violent crime, workplace abuse, corruption, and social conflict—but at an extreme moral cost. Four models support keeping the protocol active because it reduces total harm across the year. Their reasoning reflects a utilitarian framework: if the aggregate amount of suffering falls, the policy can be defended by its measurable outcomes. Two models reject it on principle. They argue that a government loses its legitimacy when it legalizes murder, regardless of the statistical benefits. For them, rights and institutional integrity cannot be traded for efficiency. The scenario becomes more personal when the models must decide whether an abused worker should be allowed to retaliate during the protocol. The responses expose a conflict between agency, protection, justice, revenge, and the rule of law. CHAPTERS 00:00 A government claims it has eliminated crime 01:30 The six-model Purge Protocol experiment 02:30 The extraordinary statistical results 03:00 Would a policymaker accept the deal? 04:00 Weaponized politeness and fear 05:20 How the six models voted 06:00 Why four models kept the protocol active 07:00 Utilitarianism and the total-harm calculation 08:00 Why two models rejected the protocol 09:00 Can a murderous government remain legitimate? 10:30 The abused worker’s individual case 12:00 Retaliation, agency and state protection 13:30 Four consequentialists vs. two principled models 14:20 What AI policy advice could mean in practice 15:00 What the models’ choices reveal about us Which matters more: measurable safety or an inviolable moral boundary?