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

    Pangram’s “100% AI” labels and the case for finer-grained scores

    A user praises Pangram’s detection abilities but argues its stark labels encourage harsh judgments and don’t show how much human work went into a text.

    MA
    1 Source, 16d ago, first seen 16d ago

    TLDR

    In one user’s experience, Pangram almost always labels each text segment “100% AI” or “100% human.” They praise its ability to distinguish AI-generated or edited text from human writing, but argue that poor calibration and a misleading interface fuel unnecessary polarization. The user says the labels don’t reveal how much human effort went into the content and wants finer-grained scores. Until then, they suggest avoiding AI for writing altogether.

    Combined views

    3K

    1 Source, first seen 16d ago

    Combined views

    3K

    1 Source, first seen 16d ago

    31 likes
    31 likes
    1 comments
    5 saves

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 comments
    5 saves

    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

    1 Source

    @maksym_andrpangram is a great tool with a clearly superhuman ability to distinguish AI-generated/edited vs. human-written text (which is clearly a big deal!). however, its lack of calibration and somewhat misleading UI leads to so much unnecessary polarization! whenever i notice the "100% AI" label, my immediate reaction is "woah, it's horrible! AI slop!". but usually it's just because the classifier is not properly calibrated (based on my experience, it almost always tells you "100% AI" or "100% human" per each segment) and doesn't tell you anything about the amount of human work went into creating the content! having more fine-grained pangram scores would've been tremendously useful. in the meantime, the best strategy seems to be avoiding using AI for writing altogether...

    1 Source

    @maksym_andrpangram is a great tool with a clearly superhuman ability to distinguish AI-generated/edited vs. human-written text (which is clearly a big deal!). however, its lack of calibration and somewhat misleading UI leads to so much unnecessary polarization! whenever i notice the "100% AI" label, my immediate reaction is "woah, it's horrible! AI slop!". but usually it's just because the classifier is not properly calibrated (based on my experience, it almost always tells you "100% AI" or "100% human" per each segment) and doesn't tell you anything about the amount of human work went into creating the content! having more fine-grained pangram scores would've been tremendously useful. in the meantime, the best strategy seems to be avoiding using AI for writing altogether...