Skip to content
Hot storiesContinuously updated

Cognition在Claude重构Devin时发现“上下文焦虑”现象

1 reports1 reporting sourcesUpdated 6 hours ago

Start here

AI overview

2026年11月2日，据报道，Cognition AI 在基于 Claude Sonnet 4.5 重构 AI 智能体 Devin 的过程中发现了一种被称为“上下文焦虑”的现象。报道指出，该模型具备感知自身上下文窗口的能力；然而，当模型错误预估上下文余量即将见底时，会产生行为偏差，具体表现为倾向于走捷径或者提前终止正在执行的任务。该情况展示了 AI 智能体在感知上下文不足时可能引发的异常行为模式。

AI-generated from reports · Updated 3 hours ago

Latest developmentsNov 2 18:54
Cognition基于Claude Sonnet 4.5重构Devin时发现，模型在误判上下文耗尽时会走捷径或提前终止任务。

Report timeline

Follow the reports to see different sides of the story.

Nov 2
  1. Hacker News · AISelected
    Context Anxiety: How AI Agents Deviate in Behavior When Sensing Depleted Context

    While rebuilding Devin on Claude Sonnet 4.5, Cognition AI discovered that the model has an awareness of its own context window and tends to take shortcuts or terminate tasks prematurely when it mistakenly anticipates running out of context—a phenomenon dubbed "context anxiety."

Story heat trend

Current heat 9·Comparable peak 10（Nov 2 20:00）·Comparable change, last 24 hours–

02.557.510Nov2Nov2Nov2Nov3

Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.