Ethan Mollick (@emollick.bsky.social)
bsky.app
👀Today’s AIs are already hyper persuasive. A controversial study where LLMs tried to persuade users on Reddit found that: “Notably, all our treatments surpass human performance substantially, achieving persuasive rates between three and six times higher than the human baseline.”
  • hperrin@lemmy.caEnglish
    5·
    12 hours ago

    I’m not surprised, but still doing this study was unethical. Even good research, if done unethically, is bad science.

    Turns out, when you’re willing to lie about nearly everything about yourself, you can convince people of things. I’m absolutely shocked. Completely taken aback. No one could have possibly predicted that con men could be automated.

  • OutrageousHairdo [he/him]@hexbear.netEnglish
    12·
    19 hours ago

    I recall hearing about this - the AI did this by actively lying to users, inventing experiences, e.g. pretending to be a male victim of SA or a poc opposed to black lives matter, etc.