LLMs are trained to block harmful responses. Old-school images can override those rules.
  • tsonfeirEnglish
    1·
    4 months ago

    People get all freaked out about these LLMs giving people directions on how to build a bomb. I’ve been using ChatGPT daily since it came out, and it’s an invaluable tool—but if I used its direction to build a bomb, I would blow my hand off.