Companies are designing AI to appear increasingly human. That can mislead users—or worse.
  • octoperson@sh.itjust.worksEnglish
    16·
    2 years ago

    Isn’t it more like, a guy who wanted to kill the Queen talked to a chatbot about it, and the chatbot rolled with it cos chatbots will default to rolling with whatever you give them? Just seems the likely scenario to me …

    Yep, this BBC article has a couple of the chat logs. If they’re representative, the bot is no more culpable than if he asked a magic 8-ball.