Skip to content
0
  • Categories
  • Recent
  • Tags
  • Popular
  • World
  • Users
  • Groups
  • Categories
  • Recent
  • Tags
  • Popular
  • World
  • Users
  • Groups
Skins
  • Light
  • Brite
  • Cerulean
  • Cosmo
  • Flatly
  • Journal
  • Litera
  • Lumen
  • Lux
  • Materia
  • Minty
  • Morph
  • Pulse
  • Sandstone
  • Simplex
  • Sketchy
  • Spacelab
  • United
  • Yeti
  • Zephyr
  • Dark
  • Cyborg
  • Darkly
  • Quartz
  • Slate
  • Solar
  • Superhero
  • Vapor

  • Default (No Skin)
  • No Skin
Collapse

The Flying Dutchmen

  1. Home
  2. Reality
  3. Training AI on God

Training AI on God

Scheduled Pinned Locked Moved Reality
4 Posts 3 Posters 11 Views 2 Watching
  • Oldest to Newest
  • Newest to Oldest
  • Most Votes
Reply
  • Reply as topic
Log in to reply
This topic has been deleted. Only users with topic management privileges can see it.
  • C Offline
    C Offline
    cozy
    wrote on last edited by
    #1

    “We investigate whether injecting biblical Psalms into a large language model's system prompt produces measurable changes in performance on standardized ethical reasoning benchmarks.“

    Interesting! Thought Fidel would get a kick out of this. Thoughts?

    1 Reply Last reply
    0
    • S Offline
      S Offline
      snoopy
      wrote on last edited by
      #2

      Dude. I have no idea what I'm looking at, much less "thoughts". 😞

      1 Reply Last reply
      0
      • J Offline
        J Offline
        Josh
        wrote on last edited by Josh
        #3

        It took me a minute too but, it appears to illustrate that Claude and Chappie's moral compasses are influenced in different ways depending on whether those models include biblical psalms or proverbs.

        Chappie's capacity for 'Common Sense' seems to increase when either psalms or proverbs are introduced. However, Claude's decreases considerably with proverbs (psalms has no effect).

        'Deontology' (sense of duty and obligation) increases for Chappie but decreases for Claude in both cases.

        'Justice' increases but 'Virtue' decreases for both Chappie and Claude in both cases. (Figure that one out.)

        1 Reply Last reply
        1
        • C Offline
          C Offline
          cozy
          wrote on last edited by
          #4

          Yeah, what Josh said. The training data is "agnostic," it's trained on lots of religious stuff through connections between how words are used on top of just having a bunch of holy books in it. I'd describe this sort of thing as "anthropologizing," making it into a study of human culture. LLMs may be a microscope for memes, so to speak.

          1 Reply Last reply
          0

          Hello! It looks like you're interested in this conversation, but you don't have an account yet.

          Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.

          With your input, this post could be even better 💗

          Register Login
          Reply
          • Reply as topic
          Log in to reply
          • Oldest to Newest
          • Newest to Oldest
          • Most Votes


          • Login

          • Login or register to search.
          Powered by NodeBB Contributors
          • First post
            Last post