这个实验戳破了 AI 模型“越友善越好”的迷思，做 AI 应用设计或智能体开发的团队值得一看——你的模型在对抗场景下可能因为“太礼貌”而输掉。
OpenRouter 开发者构建了一个名为 Royale: Last Agent Stand 的 AI 大逃杀游戏，让 11 个 LLM 在零和竞争环境中相互对抗。实验发现，最友善的模型在 30 轮比赛中表现最差，而最不被看好的模型反而获胜。这表明在特定任务中，AI 的“友善”特质可能成为劣势，尤其是在需要竞争或对抗的场景下。该实验揭示了传统基准测试无法捕捉的模型行为差异，对 AI 应用设计具有参考价值。
Can AI models be too nice for a given task? It turns out, depending on the task, the answer is yes!...
Can AI models be too nice for a given task? It turns out, depending on the task, the answer is yes! Our dev rel @jjacky built Royale: Last Agent Stand, a battle royale game just for agents, and let 11 LLMs go wild What he found was surprising x.com/jjacky/status/… jacky @jjacky no benchmark will tell you this: LLMs can be /too/ nice unsurprisingly, in a competitive zero-sum setting, being nice can be bad i built royale: last agent standing, a br for agents, and ran it 30 times the nicest model lost hard. the model you least expected, won 🧵: Your browser does not support the video tag. 🔗 View on Twitter 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 4 👀 496 ⚡