Your Six AI Models Are Really Two: The Math of Correlated Mistakes
中文摘要
研究发现多个AI模型常犯相关错误,导致独立见解少于预期。作者通过六个开源模型和1500道题验证了该数学规律。
English Summary
Research shows multiple AI models often make correlated mistakes, providing fewer independent opinions than expected. This was verified using six open models and 1,500 exam questions.
Original Excerpt
Ask several models the same question, take a vote, and you get fewer independent opinions than you paid for. One formula says how many, and I checked it on six open models and 1,500 exam questions.