观点论文：分布式后训练与去中心化对齐为何应成为一等范式
Vision Paper: Why Distributed LLM Post-Training and Decentralized Alignment Are Now Essential
一篇 arXiv 观点论文，指出 RLHF 和 DPO 默认要集中数据的设计在医疗、法律等场景根本走不通，还给出了分布式对齐的技术路线。
arXiv 上一篇观点论文主张，LLM 后训练中的分布式学习（含联邦与去中心化方式）应成为一等范式，而非仅仅是隐私附加手段。论文指出 RLHF 和 DPO 隐含假设偏好数据必须集中化，这一假设在科学上缺乏依据、实践上难以持续、伦理上有问题。作者从隐私与数据主权、多元性与代表性公平、鲁棒性与可扩展性三个支柱展开论证，并形式化了参考漂移、偏好异质性、对比损失的 DP-效用权衡等关键技术挑战。论文还提出了具体的算法方向和研究议程。
Vision Paper: Why Distributed LLM Post-Training and Decentralized Alignment Are Now Essential
This vision paper argues that distributed learning, including federated and decentralized, should become a first-class paradigm in LLM post-training, not merely a privacy add-on. Current methods like RLHF and DPO rest on an invisible assumption: preference data should be centralized. We contend this assumption is scientifically unjustified, practically unsustainable, and ethically problematic. As LLMs deploy into healthcare, law, finance, and multilingual communities, preference data originates from populations that cannot or will not centralize. We develop a three-pillar argument grounded in (1) privacy and data sovereignty, (2) pluralism and representational equity, and (3) robustness and scalability. We formalize key technical challenges unique to this intersection, reference drift, preference heterogeneity, and the DP-utility trade-off for contrastive losses, and propose concrete algorithmic directions, and sketch a research agenda. Our analysis shows distributed post-training is not only necessary but tractable, and ignoring it risks AI that is legally noncompliant, representationally narrow, and brittle at scale.
