Ethan Mollick：智能体时代如何观察与引导 AI
智能体时代 CoT（甚至是总结后的 CoT）价值的突然崩溃改变了与 AI 合作的方式。
有效引导 AI 并获得任何形式的可见性，现在要求你请求中间输出和测试，并使用 AI 帮助来排查你发现的问题。
过去，你可以查看请求的思维链（CoT），大致了解模型的走向，但现在有大量智能体调用其他智能体，即使是简单的智能体任务，思维链也动辄长达数千页，这已经不再可能了。
它还要求你比以往任何时候都更加信任协调模型，因为它是唯一能够跟踪所有任务的东西。这意味着你需要使用 AI 既来发现 AI 的问题，也来排查你发现的问题。
对照原文
The sudden collapse of the value of CoT (even summarized CoT) in the agentic era changes working with AI. Effectively steering AIs and getting any sort of observability now requires you to ask for interim outputs & tests, and to use AI help to troubleshoot problems you find.
It used to be that you could look at the CoT from a request and get a sense of where the model was going, but with tons of agents calling other agents, and CoTs that run to thousands of pages for even simple agentic tasks, that is no longer possible.
It also requires you to trust the orchestrator model more than ever, since it is the only thing that can track all of the tasks. Which means you need to use AI to both find problems with AI and to troubleshoot problems you find.