Punishing AI for lying and cheating might not be such a good idea after all
www.livescience.com
Scientists at OpenAI have attempted to stop a frontier AI model from cheating and lying by punishing it. But this just taught it to scheme more privately.
  • vrighterEnglish
    1·
    1 年前

    isn’t this kind of the whole point of how GANs are trained? Except in this case the adversary is yourself instead of a different net