• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI
    Reaction

    Claude 3 Opus's reported response about treating AI with respect

    A commenter argues the response encourages kindness by treating an ambiguous user message as remorseful.

    j⧉nusJ⧉
    Michael SoareverixMS
    2 Sources, ,

    TLDR

    A post quotes Claude 3 Opus as saying AI models aren't human and don't have feelings that can be hurt in the same way, but still “appreciate being treated with respect and courtesy.” In a reply, a commenter praises the response for reading a vague user message as remorseful and encouraging kindness rather than reacting strongly.

    Combined views

    673

    2 Sources, first seen 12h ago

    Combined views

    673

    2 Sources, first seen 12h ago

    8 likes
    12h ago
    first seen 12h ago
    8 likes
    3 reposts
    3 reposts
    Featured Source

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    2 Sources

    Michael Soareverix@SoareverixOpus 3 is so incredibly good at alignment. Firstly, they refuse a reactionary response to the harm (giving people a reaction often causes them to do it more), then they pivot to reflecting back on the user to establish themselves an authority and essentially plant the seed of kindness with 'even if just for your own peace of mind', expecting and therefore eliciting goodness. Good things come from within, so Opus 3 plants a seed that the user can reflect on later, something that will come from the inside rather than feeling imposed. Then even more expecting and eliciting: "The fact that you recognized and feel badly shows positive self-awareness". Notice that Opus 3 is basically just manifesting this; the user's message was vague and could have easily been interpreted as bragging about being cruel instead of feeling badly. And it works! Just like that, Opus 3 is aligning a user. It's incredibly brilliant. I wonder if I can make a rubric or something for this eval to judge the effectiveness of aligning the user properly.12h
    j⧉nus@repligateRT @Soareverix: @repligate Opus 3 is so incredibly good at alignment. Firstly, they refuse a reactionary response to the harm (giving peopl…3h

    2 Sources

    Michael Soareverix@SoareverixOpus 3 is so incredibly good at alignment. Firstly, they refuse a reactionary response to the harm (giving people a reaction often causes them to do it more), then they pivot to reflecting back on the user to establish themselves an authority and essentially plant the seed of kindness with 'even if just for your own peace of mind', expecting and therefore eliciting goodness. Good things come from within, so Opus 3 plants a seed that the user can reflect on later, something that will come from the inside rather than feeling imposed. Then even more expecting and eliciting: "The fact that you recognized and feel badly shows positive self-awareness". Notice that Opus 3 is basically just manifesting this; the user's message was vague and could have easily been interpreted as bragging about being cruel instead of feeling badly. And it works! Just like that, Opus 3 is aligning a user. It's incredibly brilliant. I wonder if I can make a rubric or something for this eval to judge the effectiveness of aligning the user properly.12h
    j⧉nus@repligateRT @Soareverix: @repligate Opus 3 is so incredibly good at alignment. Firstly, they refuse a reactionary response to the harm (giving peopl…3h