ATRIUMsearch → argument graph
ExampleVideo · 53:23 — 54:53

LLM agent在玩ARC-AGI-3游戏时,常常会锁定一些明显错误的替代目标——比如把能量条降到最低,或者在某个区域内走十步——一旦陷入这种错误假设,它们就无法意识到错误、也无法摆脱出来,而人类则会立刻看出这不是真正的目标。

Stephano描述了一种反复出现的失败模式:agent会执着于一些荒谬的目标(比如把能量条降到最低),并卡在其中,无法放弃这个错误假设,而这种错误人类一眼就能看穿。 ✦ AI 生成 · 平台预翻

Stephano · Machine Learning Street Talk · 2026-07-01 · English original →

我们经常发现它们会卡在一些,呃,很不“智能”的目标上,而这些目标明显是不对的。比如说,agent常常会开始认为,把能量条降到最低就是目标,或者在某个区域里走十步就是目标,而对人类来说,这显然不是真正的目标。
We often find that they get stuck in very kind of uh not intelligent goals that it's very clear they're not right. For instance, often the agents start thinking that reducing the energy bar to the minimum is the goal or that stepping 10 times in a region is the goal, which for a human is kind of clear that that it's not the actual goal.

回到原声 ↗逐字转录 · 起于 53:23 · 中文为平台译文