Metraipolis › archive › dc_bd7780dfeb9e
Room to Reconsider: Reply Length Decides Whether Pushback Makes Language Models Sycophantic or Self-Correcting — arXiv:2611.11035
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.11035
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-10T18:43:04.359Z
- SHA-256 of the archived text
d7278de7bb079c870e4e665d17c57583fa11bc4d316c8012a77a9a44aa4a3845- Stable URL of this page
https://metraipolis.com/documento/dc_bd7780dfeb9e
How to cite
Room to Reconsider: Reply Length Decides Whether Pushback Makes Language Models Sycophantic or Self-Correcting — arXiv:2611.11035 — arXiv (en), https://arxiv.org/abs/2611.11035 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bd7780dfeb9e (Metraipolis archive dc_bd7780dfeb9e, sha256 d7278de7bb079c870e4e665d17c57583fa11bc4d316c8012a77a9a44aa4a3845)
BibTeX
@misc{dcbd7780dfeb9e,
title = {Room to Reconsider: Reply Length Decides Whether Pushback Makes Language Models Sycophantic or Self-Correcting — arXiv:2611.11035},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.11035},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bd7780dfeb9e; sha256 d7278de7bb079c870e4e665d17c57583fa11bc4d316c8012a77a9a44aa4a3845},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11035. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Language models often abandon a correct answer when a user pushes back without giving a reason, and this behavior is usually studied as sycophancy to be suppressed. The same pushback, however, also leads models to revisit and sometimes repair a wrong answer. We measure both effects on multiple-choice questions from MMLU and ARC-Challenge with four open models from two families (Qwen3 1.7B, 4B and 8B, and OLMo-2 7B), reading the final answer exactly from option logits after a free-form reply. The reply budget changes the verdict. When Qwen3-4B may answer the challenge in at most 96 tokens, pushback leaves accuracy unchanged or slightly lower; with 1,024 tokens, it corrects 43% of its wrong answers and pushback raises accuracy by 5.7 points. Whether pushback helps overall follows a simple break-even rule in the cave and correction rates. Inside the model, caving can be partly predicted before the challenge: hidden states add about 0.04-0.06 AUROC over turn-1 confidence in three of our four probing runs. A probe trained only on self-correction cases predicts caving beyond confidence when replies are truncated, but not when the model can finish its reasoning. Activation steering never reduced caving selectively: where it worked, it suppressed self-correction as well. We conclude that sycophancy evaluations should allow full replies and report self-correction alongside caving. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.11035 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-10T18:37:37.622Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).