• Pennomi@lemmy.world
      link
      fedilink
      English
      arrow-up
      73
      ·
      8 months ago

      We optimize our AI by human preference (RLHF). People don’t seem to understand that this will lead to every problem we’ve seen in social media - an endless stream of content that engages you addictively.

      The AI uprising will be slow and gentle, and we will welcome it.