Imagine every report has the following: Agent's best-guess about what comments you'd get from Beth, Hjalmar, Ajeya. Agent's best-guess about survey results. Agent's best-guess about benchmark results. Agent's best-guess about how this will be received on Twitter.
「预测反馈」的概念令人惊讶：AI 在报告发出前，预测各位审阅者会说什么、Twitter 会怎么反应、调查结果会是什么——研究者先在「预测反馈」中迭代，只有当预期信息增量足够高时，才真正发出去等待真实反馈。这是一种「反馈的预计算」——把等待时间转化为优化时间，本质上是把「串行等待」变成了「并行模拟」。