arXiv cs.CL· Letian Shi, Kanwal Mehreen, Deheng Zhang, Yuxia Wang·· 12 小时前AI 评分31
研究：共情并非模型事实性与可靠性下降的替罪羊
Empathy Is Not the Scapegoat for Declines in Factuality and Reliability
AI 导读
针对“更温暖的模型可靠性更低”的既有结论，该研究指出其并不完整：过往实验多在共情对话上训练，难以区分语气效应与 SFT 带来的分布偏移。作者从提示词、SFT 和 RL 三种方式诱导温暖行为，用 11 个数据集基准套件和情感语境下的 EMOFACT 评估能力保持情况，发现温暖效应因任务和模型而异，任务对齐的自适应语气可同时提升共情与特定可靠性，但仍存在部分能力权衡。研究还提出经人工验证的共情评估器。
正文 · 原文
Abstract:Language models are increasingly expected to support users in stressful or uncertain situations, where warmth and empathy are desirable. While recent work associates warmer models with reduced reliability, we show that this conclusion is incomplete. Prior studies often train primarily on empathetic dialogue, making it difficult to separate tone effects from SFT-induced distribution shift. We revisit this issue at both the model and evaluation levels. We study prompting, SFT, and RL for inducing warm behavior, and comprehensively evaluate capability preservation using an 11-dataset benchmark suite and EMOFACT under emotional contexts. Across multiple model families, we find that warmth effects are task- and model-dependent, while task-aligned adaptive tone can improve empathy and targeted reliability, although some capability trade-offs remain. We further introduce human-validated empathy evaluators specifically developed to assess empathetic model responses. Our results suggest that models can be more supportive in social interactions while remaining reliable.
| Comments: | Accepted by AACL 2026 |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.18541 [cs.CL] |
| (or arXiv:2610.18541v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.18541 arXiv-issued DOI via DataCite |
Submission history
From: Letian Shi [view email]
[v1]
Tue, 13 Oct 2026 15:20:46 UTC (160 KB)
[v2]
Mon, 2 Nov 2026 14:33:43 UTC (152 KB)
来源：arXiv cs.CL · arxiv.org