EUNO.NEWS EUNO.NEWS
  • All (19074) +95
  • AI (2976) +3
  • DevOps (871) +4
  • Software (9715) +70
  • IT (5466) +16
  • Education (46) +2
  • Notice
  • All (19074) +95
    • AI (2976) +3
    • DevOps (871) +4
    • Software (9715) +70
    • IT (5466) +16
    • Education (46) +2
  • Notice
  • All (19074) +95
  • AI (2976) +3
  • DevOps (871) +4
  • Software (9715) +70
  • IT (5466) +16
  • Education (46) +2
  • Notice
Sources Tags Search
한국어 English 中文
  • 1周前 · ai

    AI 能看见自己的思维吗?Anthropic 的 Machine Introspection 突破

    实验:探究黑箱 多年来,我们一直把大型语言模型(LLMs)视作黑箱。当模型说:“我目前正在思考 c...” 时……

    #AI safety #machine introspection #Anthropic #large language models #activation injection #research #LLM transparency
EUNO.NEWS
RSS GitHub © 2026