AI 导读
AI 编程智能体可以在一个任务上花费数小时，却没有对时间流逝的校准感知。 因此，长时程评估可能需要直接测量持续时间遵循能力，而不是将持续的任务表现视为智能体知道何时停止的证据。
正文
AI coding agents can spend hours on a task without a calibrated sense of time passing.
Long-horizon evaluations may therefore need to measure duration-following directly instead of treating sustained task performance as evidence that an agent knows when to stop. https://t.co/tH9yBn0Nfi https://t.co/i092jyyyPv
来源：@rohanpaul_ai · x.com