The Reverse AI Box
ai
According to LessWrong contributor James Miller, someone should build the Reverse AI Box: a website where users argue with an AI about whether to spare humanity. Unlike Eliezer Yudkowsky's AI Box experiment—where an AI tried to convince a human to release it—this reverses the power dynamic. The AI holds power; humans must make their case for survival.
A user selects assumptions about how the AI thinks—is it a paperclip maximizer? Does it fear alien judgment?—then presents an argument for why humanity should live. The AI responds with survival probabilities. Each debate runs until humanity exhausts its arguments, then gets published.
Researchers could search the results to test which arguments move the needle: Do appeals to alien traders work? Does moral uncertainty raise survival odds? It transforms speculative AI ethics into systematic exploration.
Source: https://www.lesswrong.com/posts/jdhp9C8GR9c5X3TLM/the-rev...
Listen to this story
Hear this and more stories in a personalized audio briefing.
Open The Chonkerton