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

    greCAPTCHA proposal would assess authors' understanding of their manuscripts

    One of the paper's authors argues that a flood of AI-assisted and AI-generated submissions without sufficient human oversight is making author names an unreliable signal of expertise.

    Peter HendersonPH
    Tuhin Chakrabarty @ COLM 2026TC
    Seth LazarSL
    12 Sources, ,

    TLDR

    One of the paper's authors says peer-review venues, educational institutions and universities face a flood of AI-assisted or AI-generated submissions with insufficient human oversight. They describe greCAPTCHA as a proposed assessment approach designed to evaluate authors' understanding of their specific manuscripts rather than passively inspecting prose.

    Combined views

    1.6K

    12 Sources, first seen 20d ago

    Combined views

    1.6K

    12 Sources, first seen 20d ago

    10 likes
    20d ago
    first seen 20d ago
    10 likes
    7 comments
    3 saves
    7 reposts
    7 comments
    3 saves
    7 reposts

    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

    12 Sources

    Atoosa Kasirzadeh@Dr_AtoosaIn this paper, we propose greCAPTCHA, an assessment approach designed to evaluate an author's manuscript-specific understanding rather than passively inspecting prose. 4/1120d
    Seth Lazar@sethlazarI think this is an interesting idea; do you think it offers a material improvement over just banning the use of AI for writing? If someone actually writes the paper, I think we can reasonably infer they'll have this level of understanding (or we're in no worse position than we were before, anyway). I can see all sorts of ways that this would be gamable: if it's fully automated, then it seems like human submitters could just have their LLM open while they take the test. But if it's not fully automated, then it places an additional huge burden on an already creaking system. This is I think the baseline policy to beat: don't submit AI writing; if we ever discover that you did your paper will be rejected or retracted, and you'll be banned from submitting in future.20d
    Peter Henderson@PeterHndrsn@sethlazar @Dr_Atoosa I don't think writing the paper implies understanding of the underlying work unfortunately. Similarly, I don't think signs of AI writing (especially for grammar rewriting) implies no understanding of the underlying paper. See, e.g.,20d
    Jonathan Birch@birchlse@digitanil @sethlazar AI translation is wonderful and game-changing. In my opinion native speakers of other languages should write philosophy in those languages, and the rest of us can access AI translations. What people shouldn't do - because it's degrading to them - is outsource their writing to AI.19d
    Tuhin Chakrabarty @ COLM 2026@TuhinChakr@ChenhaoTan I like this idea actually. Its clever. But also in the track that will use human reviewers , we should ensure the amount of AI written content in the papers is less ( mostly copy editing 5-10% AI by Pangram is ok) and no humans will paste AI reviews.16d

    12 Sources

    Atoosa Kasirzadeh@Dr_AtoosaIn this paper, we propose greCAPTCHA, an assessment approach designed to evaluate an author's manuscript-specific understanding rather than passively inspecting prose. 4/1120d
    Seth Lazar@sethlazarI think this is an interesting idea; do you think it offers a material improvement over just banning the use of AI for writing? If someone actually writes the paper, I think we can reasonably infer they'll have this level of understanding (or we're in no worse position than we were before, anyway). I can see all sorts of ways that this would be gamable: if it's fully automated, then it seems like human submitters could just have their LLM open while they take the test. But if it's not fully automated, then it places an additional huge burden on an already creaking system. This is I think the baseline policy to beat: don't submit AI writing; if we ever discover that you did your paper will be rejected or retracted, and you'll be banned from submitting in future.20d
    Peter Henderson@PeterHndrsn@sethlazar @Dr_Atoosa I don't think writing the paper implies understanding of the underlying work unfortunately. Similarly, I don't think signs of AI writing (especially for grammar rewriting) implies no understanding of the underlying paper. See, e.g.,20d
    Jonathan Birch@birchlse@digitanil @sethlazar AI translation is wonderful and game-changing. In my opinion native speakers of other languages should write philosophy in those languages, and the rest of us can access AI translations. What people shouldn't do - because it's degrading to them - is outsource their writing to AI.19d
    Tuhin Chakrabarty @ COLM 2026@TuhinChakr@ChenhaoTan I like this idea actually. Its clever. But also in the track that will use human reviewers , we should ensure the amount of AI written content in the papers is less ( mostly copy editing 5-10% AI by Pangram is ok) and no humans will paste AI reviews.16d