Unverified50% confidenceFactExact time
算法厌恶指人们对算法犯错的容忍度远低于对人类犯错的容忍度
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
Unverifiedtyposquatting依赖人为错误，而slopsquatting依赖AI错误，后者更危险因为AI错误是系统性且可预测的70% similarUnverified模型自动路由算法对'最优模型'的判断标准往往与用户主观偏好存在错位，当算法优先考虑成本或延迟时可能牺牲输出质量68% similarUnverified过度严苛的内容限制可能扼杀正当的创作自由，演变为算法审查68% similarUnverified垃圾内容制造者已开始刻意制造"不完美"（如全小写、无标点、拼写错误）来逃避AI检测机制68% similarUnverified算法优化的是概率最优解而非美学最优解，AI译本容易产生平庸而缺乏灵魂的译句67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/895042API
curl https://kongchang.com/api/v1/knowledge/claims/895042MCP
get_claim(id=895042)