Skip to content
AI News HubLIVE
Original source2 min read

Benchmarks Don't Talk Back: Learning Through Self-Disagreement

Summary

An arXiv paper introduces "dialectical competence"—the ability to know when to hold or yield a position—as a capability distinct from accuracy, and trains it through self-generated antitheses with verifiable rewards, requiring no second model, reward model, or human annotation.

SourcearXiv Computational LinguisticsAuthor: Soheil Abbasloo, Peng Cheng
Benchmarks Don't Talk Back: Learning Through Self-Disagreement
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 12 Nov 2026]

Title:Benchmarks Don't Talk Back: Learning Through Self-Disagreement

View a PDF of the paper titled Benchmarks Don't Talk Back: Learning Through Self-Disagreement, by Soheil Abbasloo and 1 other authors

View PDF HTML (experimental)

Abstract:A benchmark usually measures a language model alone with a question, and deployment rarely looks like that: answers get contested, and models give up correct ones under counter-reasoning that is itself wrong. We show that what fails there is not a shortfall of accuracy but a capability in its own right -- knowing when to hold a position and when to yield to a better one -- which we call dialectical competence. It stands on first-answer accuracy without following from it: a model can be made materially better at answering and no better at this, and training for accuracy alone can drift against it. Nor can it be bought at inference, by self-reflection or by sampling more. Taking a cue from the dialectical tradition in philosophy, in which productive opposition arises within a position rather than being imposed on it from outside, we hypothesize that this capability can be trained from opposition the model supplies itself. It states a position, generates the strongest case against that position -- its antithesis -- rules on whether the case stands, and is trained on that ruling, with no second model at training or at deployment. We call this learning through Self-Disagreement: reinforcement learning from verifiable rewards throughout, where the same check that scores an answer against its reference scores the ruling too, so no reward model, no preference data and no human annotation enter anywhere. The capability is trained rather than elicited: weak in the base model, it arrives with the reward, is not paid for out of first-answer accuracy, and holds against challengers the model never trained against, including ones stronger than itself. The opposition a model needs in order to improve, it turns out, need not come from outside it.

Comments: 34 pages, 5 figures, 15 tables

Subjects:

Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)

Cite as: arXiv:2611.17563 [cs.CL]

(or arXiv:2611.17563v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2611.17563

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Soheil Abbasloo [view email] [v1] Thu, 12 Nov 2026 22:37:32 UTC (227 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Benchmarks Don't Talk Back: Learning Through Self-Disagreement, by Soheil Abbasloo and 1 other authors

View PDF

HTML (experimental)

TeX Source

view license

Additional Features

Audio Summary

Current browse context:

cs.CL

new | recent | 2026-11

Change to browse by:

cs cs.AI cs.LG

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • Benchmarks measure models answering alone, but deployment involves contested answers, and models can abandon correct answers under faulty counter-reasoning.
  • The authors define dialectical competence as knowing when to hold a position and when to yield, which does not follow from first-answer accuracy and can degrade under accuracy-only training.
  • The method, learning through Self-Disagreement, has the model state a position, generate its strongest antithesis, rule on it, and train on that ruling.
  • Everything uses reinforcement learning from verifiable rewards, with no reward model, preference data, or human annotation, and the capability generalizes to stronger unseen challengers.

Highlights and analysis are generated automatically and may contain errors. Check the original source.