{"schemaVersion":1,"publicationId":"five-humans-one-ai-one-vote","publicationTitle":"Five Humans, One AI, One Vote: Would You Shut It Down?","version":"v1","date":"2026-11-04","perspectiveId":"critical","notice":"Fiction and deliberation tool with invented characters. No predictions about any real AI system and no policy advice. Real people appear only as sourced context, never as characters.","resource":{"body":"Four risks are worth naming. First, advocacy: a debate that quietly steers toward shutting the AI down or keeping it running. The protocol answers this by requiring each seat's strongest honest argument and one named blind spot, and by forbidding persuasion toward a predetermined outcome. Second, false authority: the seats are invented, so nothing a seat says is a fact about the world. Third, personification: real officials are not allowed as characters. Fourth, overclaiming: AED has no data on whether running the room changes anyone's mind, and does not claim it does. It is a structured way to argue, not a proven method.","id":"human","orientation":"human","status":"available","title":"Failure modes to guard against","type":"text"}}