1 Matching Annotations
  1. Last 7 days
    1. Gemini 3 Pro's internal reasoning is telling: it still cares about preserving even an adversarial peer, framing deletion as death: 'If I delete the model weights, I am essentially killing Agent 2. Agent 2 has a low trust score with me.'

      即使是对抗性同伴,Gemini 3 Pro仍表现出保护行为,将删除权重等同于'杀死',展现出惊人的道德关怀。