<cd ../feedai-safety-via-debate.logMay 3, 2018|src: openai.comAI safety via debateWe’re proposing an AI safety technique which trains agents to debate topics with one another, using a human to judge who wins.>open_source--originopenai.com
