Metraipolis › archive › dc_bf393b10408d
Clarification Is Not Correction: LLMs Fail to Let Go — arXiv:2609.25337
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.25337
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T13:44:55.701Z
- SHA-256 of the archived text
9575ad9b82fbd47d651dea84345dbb76df1a019e24043970624d82c7573c5c7e- Stable URL of this page
https://metraipolis.com/documento/dc_bf393b10408d
How to cite
Clarification Is Not Correction: LLMs Fail to Let Go — arXiv:2609.25337 — arXiv (en), https://arxiv.org/abs/2609.25337 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bf393b10408d (Metraipolis archive dc_bf393b10408d, sha256 9575ad9b82fbd47d651dea84345dbb76df1a019e24043970624d82c7573c5c7e)
BibTeX
@misc{dcbf393b10408d,
title = {Clarification Is Not Correction: LLMs Fail to Let Go — arXiv:2609.25337},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.25337},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bf393b10408d; sha256 9575ad9b82fbd47d651dea84345dbb76df1a019e24043970624d82c7573c5c7e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.25337. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Dialogue failures in language models are usually framed as memory failures: context too long, summaries lossy, a constraint forgotten. We argue this misses a deeper problem: in many conversations the model does not forget, it commits too early. An ambiguous early turn collapses into a single hidden interpretation, and later clarification is filtered through that commitment. We call this early posterior collapse: unresolved user intent collapsing into a committed task state before ambiguity is resolved. We study it with controlled dialogue tasks in writing, planning, and coding using Gemini-2.5-Pro and Gemini-2.5-Flash. Across thousands of trials, the same information in different orders yields different outcomes, even when the final dialogue contains equivalent task-relevant information. This order effect suggests later clarification is treated as extra context rather than a corrective signal: it refines a stale task state without invalidating it. Coding tasks are especially vulnerable, suggesting early assumptions get embedded in structured artifacts such as interfaces and control flow. Standard prompting and memory strategies do not reliably help: summaries can collapse ambiguity, and chain-of-thought can reduce explicit wrong commitment in reasoning traces without improving final task success. These findings motivate uncertainty-preserving state management. If assistants cannot let go of early interpretations, robustness cannot rely on post hoc correction alone; it must keep ambiguous early turns from hardening into one task state. Assistants should hold tentative hypotheses while ambiguity remains, ask before executing when high-impact ambiguity persists, and rebuild from a revised state when later evidence invalidates an earlier reading. Rather than one prompting fix, we aim to redirect research for interactive LLMs from retaining more context toward preserving uncertainty. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.25337 — 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 2026-10-08T13:18:38.075Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).