4 pointsby tattattaei7 hours ago1 comment
  • AnimalMuppet6 hours ago
    It may be a positive feedback loop - whichever language the AIs write more of becomes more predominant in future training data.

    But if AIs write bugs in that language, they also poison the well...

    • tattattaei6 hours ago
      Model collapse of the winning language