Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified50% confidenceFactExact time

绝大多数RLHF标注工作以英语为主，导致不同语言间对齐质量存在结构性失衡

1

Sources

50%

Confidence

Long-term

Relevance

7/16/2026

First Seen

Sources

AI多语言偏差：Claude在印地语和阿拉伯语中更"礼貌"的背后

Related Claims

UnverifiedAnthropic、OpenAI等机构的对齐标注工作大量依赖英语母语标注员，绝大多数RLHF标注工作以英语为主78% similar Unverified许多模型的安全对齐在非英语语言中表现显著下降，低资源语言尤其容易被利用，因为RLHF对齐训练的人类标注数据高度集中在英语上76% similar UnverifiedRLHF标注工作对标注者要求远高于传统众包任务，需要具备较强的语言理解能力、领域知识乃至编程能力71% similar Unverified英式发音（RP）与美式发音选一种为主，避免同一套读物内英美混杂，早期混杂会干扰儿童音素辨识；后续可作对比拓展66% similar UnverifiedOpenAI的RLHF和Anthropic的Constitutional AI等主流对齐方法本质上包含对输出语言风格的隐性约束65% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/532880

API

curl https://kongchang.com/api/v1/knowledge/claims/532880

MCP

get_claim(id=532880)