Event ·Multi-source

OpenAI 莉莉计划人工审核 ChatGPT 聊天记录优化模型

OpenAI 内部评估项目“莉莉计划”让“提示词审核员”查看匿名化用户聊天记录，评判 ChatGPT 回复质量，检查是否出现“AI 式话术”或“居高临下的说教语气”。审核人员有时会接触到用户私人信息，但 OpenAI 承认过滤环节可能漏掉部分个人数据。该计划不专门核查回答准确性，仅标记明显错误，与排查用户伤害意图的人工安全审查是独立流程。

Current read

OpenAI 内部项目曝光，讲清楚他们怎么用人工审核优化 ChatGPT，还提到审核员时薪超过 50 美元，挺有意思的。

11 sources78° AI heatLast updated 2026/9/15 11:48:13

Evidence

11 sources

Primary sourceIT之家

2026/9/15 11:48:13View source

Related source 1@OpenAIDevs

2026/9/15 20:11:45View source

Related source 2VnExpress Số hóa

2026/9/16 00:52:22View source

Related source 3宝玉

2026/9/16 06:30:01View source

Related source 4小互

2026/9/14 00:32:03View source

Related source 5CNBC Indonesia Tech

2026/9/14 00:35:00View source

Related source 6AI Will

2026/9/14 01:07:27View source

Related source 7Techsauce

2026/9/14 06:16:24View source

Related source 8量子位

2026/9/14 09:36:28View source

Related source 9Gary Marcus

2026/9/14 13:06:31View source

Related source 10a16z

2026/9/14 14:32:27View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story