Hao AI Lab：GPT6-Astra 领队 Jev 集群破案测试
4️⃣ But when the witnesses are bribed… 💰
AI 导读
Hao AI Lab 用 200 个维多利亚伦敦探案案例测试智能体集群：在 100 个较难、付费证人多于诚实证人的案例中，GPT6-Astra 带领 Jev 集群解决 93 例，GPT6-Astra 单独 59 例，Jev 集群单独 42 例，规则式领导者加 Jev 集群仅 0–1 例。作者结论是：面对有组织的谎言，需要一个会追问“谁在说话”的推理型领导者。
正文
4️⃣ But when the witnesses are bribed… 💰
In 100 harder cases, paid witnesses outnumber honest ones. Here is how each team performs: • 🏅 GPT6-Astra + Jev swarm: 93 • 🥈 GPT6-Astra alone: 59 • 🥉 Jev swarm alone: 42 • ❓ Rule-based leader + Jev swarm: 0–1
Takeaway: against organized lies, you need a reasoning leader who asks who's talking.