AI Agent 上下文获取：主动推断与最优提问
What a fascinating paper on AI agents.
Agent 领域的新范式，用主动推断解决上下文缺失痛点，有明确的数学定义和基准测试，值得深入研读。
What a fascinating paper on AI agents.
这篇关于 AI 智能体的论文真引人入胜。
A lot of the issues we see with AI agents today revolve around wrong assumptions the LLMs make.
如今我们在 AI 智能体方面遇到的许多问题，都源于大语言模型做出的错误假设。
This leads to problems like hallucination, cost inefficiencies, unreliable tool calls and much more.
这导致了幻觉、成本效率低下、不可靠的工具调用等种种问题。
I think if we can solve this problem, even current LLMs would significantly improve in terms of performance and efficiency.
我认为如果我们能解决这个问题，即使是当前的大语言模型，其性能和效率也会显著提升。
The problem is that context acquisition is treated as afterthought, but it shouldn't be that way.
问题在于，上下文获取往往被视为次要事项，但这本不应如此。
Users tend to leave out constraints when prompting. So the agent agent needs to guess the default, or spend tokens on a clarifying question, a retrieval call, a tool call, or a prompt trial.
用户在提示时往往会遗漏约束条件。因此，智能体需要猜测默认值，或者花费 token 去提出澄清问题、进行检索调用、工具调用或尝试不同的提示词。
This new work gives this problem an objective function. Context acquisition becomes active inference over a latent task state. An inner step updates beliefs, and an outer step picks the next context action, task action, or stop action to minimize expected free energy under cost.
这项新工作为这个问题提供了一个目标函数。上下文获取变成了对潜在任务状态的主动推断。一个内部步骤更新信念，一个外部步骤选择下一个上下文动作、任务动作或停止动作，以在成本约束下最小化期望自由能。
In deterministic settings the epistemic term reduces to expected information gain, optionally normalized by token cost. That is directly implementable today as a scoring rule.
在确定性设置中，认知项简化为期望信息增益，可选择性地按 token 成本进行归一化。这可以直接作为评分规则在今天实现。
They coin it as Optimal Question Asking, with exact posteriors and a dynamic programming oracle, then benchmark frontier models on binary and multiway tasks from 25 to 300 candidates. So you can measure the gap between your agent and the true optimum.
他们将其命名为“最优提问”，使用精确的后验概率和动态规划预言机，然后在包含 25 到 300 个候选项的二元和多路任务上对前沿模型进行基准测试。这样你就可以衡量你的智能体与真正最优解之间的差距。
Paper: https://arxiv.org/abs/2608.19202
论文：https://arxiv.org/abs/2608.19202
Track more trending AI papers in our academy: https://academy.dair.ai/
在我们的学院中追踪更多热门 AI 论文：https://academy.dair.ai/
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力