• 17 Posts
  • 136 Comments
Joined 3 年前
cake
Cake day: 2023年7月3日

help-circle

















  • I think most of us can agree that, ethical concerns about training data or energy cost aside, some use cases of LLMs are unproblematic given that you are actually validating the output. Meeting summaries, text manipulation, etc.

    The real problem here is that Kristersson seems to have used it in more dubious ways, “asking it for advice”. Depending on the specific LLM used this makes his use rank anywhere between “somewhat problematic” and “catastrophic”.