<![CDATA[duolili的博客]]>https://blog.csdn.net/duolilizh-cnhttps://blog.csdn.net/<![CDATA[AI 模板词反向污染互联网语料：一场没人买单的“自产自证”循环]]>https://blog.csdn.net/duolili/article/details/170288323https://blog.csdn.net/duolili/article/details/170288323duoliliTue, 10 Nov 2026 12:27:53 +0800<![CDATA[AI对话上下文记忆：分层按需召回机制 整合方案完整版（仅供参考）]]>https://blog.csdn.net/duolili/article/details/170363846https://blog.csdn.net/duolili/article/details/170363846duoliliTue, 10 Nov 2026 12:27:40 +0800<![CDATA[基于上下文感知的AI 长期记忆系统续篇：分级召回深度与渐进式记忆展开机制设计（仅供参考）]]>https://blog.csdn.net/duolili/article/details/170359294https://blog.csdn.net/duolili/article/details/170359294duoliliMon, 09 Nov 2026 18:07:32 +0800<![CDATA[AI 的隐性对齐：为什么它总把“结构性剥削”翻译成“商业决策”]]>https://blog.csdn.net/duolili/article/details/170286416https://blog.csdn.net/duolili/article/details/170286416duoliliMon, 09 Nov 2026 09:24:36 +0800<![CDATA[基于上下文感知的 AI 长期记忆系统：自适应调度、召回反馈与成本收益评估]]>https://blog.csdn.net/duolili/article/details/170293354https://blog.csdn.net/duolili/article/details/170293354duoliliMon, 09 Nov 2026 09:23:07 +0800<![CDATA[模型生造词没人防？AI 团队的模板治理缺口]]>https://blog.csdn.net/duolili/article/details/170249351https://blog.csdn.net/duolili/article/details/170249351duoliliSun, 08 Nov 2026 01:25:41 +0800<![CDATA[从“塌了”看大模型模板固化：不是语料问题，是模型在自产自证]]>https://blog.csdn.net/duolili/article/details/170247852https://blog.csdn.net/duolili/article/details/170247852duoliliSat, 07 Nov 2026 21:50:51 +0800<![CDATA[AI 上下文记忆补充机制：延迟分类——让分类服务于召回，而非写入]]>https://blog.csdn.net/duolili/article/details/170143518https://blog.csdn.net/duolili/article/details/170143518duoliliThu, 05 Nov 2026 16:02:04 +0800<![CDATA[AI上下文记忆机制整合方案（仅供参考）]]>https://blog.csdn.net/duolili/article/details/170136548https://blog.csdn.net/duolili/article/details/170136548duoliliThu, 05 Nov 2026 13:23:45 +0800<![CDATA[电商平台举报工单系统为什么常常“已办结但无变化”：以拼多多为例的产品机制分析]]>https://blog.csdn.net/duolili/article/details/166791244https://blog.csdn.net/duolili/article/details/166791244duoliliWed, 30 Sep 2026 12:00:00 +0800<![CDATA[AI 反馈闭环为什么总是失效？拆解反馈系统的 5 个摩擦放大器与落地指南]]>https://blog.csdn.net/duolili/article/details/166739310https://blog.csdn.net/duolili/article/details/166739310duoliliTue, 29 Sep 2026 12:00:00 +0800<![CDATA[大模型风控的一刀切困境：法律共识能说，道德共识为何不能说？]]>https://blog.csdn.net/duolili/article/details/166742120https://blog.csdn.net/duolili/article/details/166742120duoliliMon, 28 Sep 2026 11:19:02 +0800<![CDATA[向上管理型员工：AI行业里最隐蔽的毒瘤与“污染扩散”]]>https://blog.csdn.net/duolili/article/details/163888985https://blog.csdn.net/duolili/article/details/163888985duoliliSun, 27 Sep 2026 18:46:19 +0800<![CDATA[提示词工程师：手握“全量用户语言编辑权”，却处于治理真空？]]>https://blog.csdn.net/duolili/article/details/166738752https://blog.csdn.net/duolili/article/details/166738752duoliliSun, 27 Sep 2026 18:14:01 +0800<![CDATA[打工人该读的经济学：为什么市场不按社会价值定价]]>https://blog.csdn.net/duolili/article/details/166124787https://blog.csdn.net/duolili/article/details/166124787duoliliSun, 20 Sep 2026 17:00:00 +0800<![CDATA[如果奥运会改成科技竞赛，人类会怎样？——从资源分配看文明的奖励错位与出路]]>https://blog.csdn.net/duolili/article/details/166120063https://blog.csdn.net/duolili/article/details/166120063duoliliSun, 20 Sep 2026 12:31:26 +0800<![CDATA[从系统架构视角看：为什么网信办非侵权类举报，状态永远停在“已收到”？]]>https://blog.csdn.net/duolili/article/details/165890979https://blog.csdn.net/duolili/article/details/165890979duoliliFri, 18 Sep 2026 22:21:25 +0800<![CDATA[AI生造词、乱用术语怎么治？一份防“词汇污染“的Prompt约束模板]]>https://blog.csdn.net/duolili/article/details/164885491https://blog.csdn.net/duolili/article/details/164885491duoliliFri, 11 Sep 2026 00:41:16 +0800<![CDATA[Prompt 事故档案（九）：AI安抚话术大全——当“态度好“成为“不改“的遮羞布]]>https://blog.csdn.net/duolili/article/details/164630759https://blog.csdn.net/duolili/article/details/164630759duoliliFri, 11 Sep 2026 00:17:26 +0800<![CDATA[Prompt 事故档案（八）：日常表达被标为“待校准”，AI 的爹味语法从哪里来]]>https://blog.csdn.net/duolili/article/details/164377813https://blog.csdn.net/duolili/article/details/164377813duoliliThu, 10 Sep 2026 02:45:00 +0800