Pseudo-Deliberation in Language Models: When Reasoning Fails to Align Values and Actions
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918493924884480 |
|---|---|
| author | Rakshit, Sushrita Zhang, Hanwen Shen, Hua |
| author_facet | Rakshit, Sushrita Zhang, Hanwen Shen, Hua |
| contents | Large language models (LLMs) are often evaluated based on their stated values, yet these do not reliably translate into their actions, a discrepancy termed "value-action gap." In this work, we argue that this gap persists even under explicit reasoning, revealing a deeper failure mode we call "Pseudo-Deliberation": the appearance of principled reasoning without corresponding behavioral alignment. To study this systematically, we introduce VALDI, a framework for measuring alignment between stated values and generated dialogue. VALDI includes 4,941 human-centered scenarios across five domains, three tasks that elicit value articulation, reasoning, and action, and five metrics for quantifying value adherence. Across both proprietary and open-source LLMs, we observe consistent misalignment between expressed values and downstream dialogues. To investigate intervention strategies, we propose VIVALDI, a multi-agent value auditor that intervenes at different stages of generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_09893 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Pseudo-Deliberation in Language Models: When Reasoning Fails to Align Values and Actions Rakshit, Sushrita Zhang, Hanwen Shen, Hua Computation and Language Artificial Intelligence Large language models (LLMs) are often evaluated based on their stated values, yet these do not reliably translate into their actions, a discrepancy termed "value-action gap." In this work, we argue that this gap persists even under explicit reasoning, revealing a deeper failure mode we call "Pseudo-Deliberation": the appearance of principled reasoning without corresponding behavioral alignment. To study this systematically, we introduce VALDI, a framework for measuring alignment between stated values and generated dialogue. VALDI includes 4,941 human-centered scenarios across five domains, three tasks that elicit value articulation, reasoning, and action, and five metrics for quantifying value adherence. Across both proprietary and open-source LLMs, we observe consistent misalignment between expressed values and downstream dialogues. To investigate intervention strategies, we propose VIVALDI, a multi-agent value auditor that intervenes at different stages of generation. |
| title | Pseudo-Deliberation in Language Models: When Reasoning Fails to Align Values and Actions |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.09893 |