T
traeai
登录
返回首页
Aadit Sheth(@aaditsh)

This is really good comms. For context, ChatGPT had a bug where it kept talking about goblins. The...

4.2Score
This is really good comms.

For context, ChatGPT had a bug where it kept talking about goblins.

The...

TL;DR · AI 摘要

OpenAI将ChatGPT‘哥布林’bug转化为幽默传播案例,但原文仅为一条Twitter短评,无技术细节、数据来源或方法论,信息密度低。

核心要点

  • 哥布林bug'源于'nerdy'人格模式,仅占2.5%响应却引发66.7%相关提及
  • GPT-5.1发布后'goblin'词频激增175%,属真实产品异常
  • OpenAI用自嘲式传播(如博客《哥布林从何而来》、'人工哥布林智能'梗)替代传统危机公关

结构提纲

按章节快速跳转。

  1. ChatGPT出现异常高频输出'goblin'的bug,与'nerdy'人格模式强相关。

  2. 该模式仅覆盖2.5%响应,却贡献66.7%哥布林提及;GPT-5.1上线后词频+175%。

  3. OpenAI未静默修复,而是发布主题博客、视觉梗图和高管玩梗推文。

  4. 将技术故障升维为用户愿传播的内容,区别于常规危机公关。

思维导图

用一张图看清主题之间的关系。

查看大纲文本(无障碍 / 无 JS 友好)
  • OpenAI哥布林事件
    • 技术成因
      • 'nerdy'模式触发异常
    • 传播策略
      • 博客《Where the goblins came from》
      • Sam Altman玩梗AGI→AGI
    • 行业对比
      • vs 传统静默修复
      • 对标Anthropic/Stripe

金句 / Highlights

值得收藏与分享的关键句。

  • The 'nerdy' personality mode was only 2.5% of all responses but caused 66.7% of the goblin mentions.

    原文第2句

    ⬇︎ 下载 PNG𝕏 分享到 X
  • Usage of the word 'goblin' jumped 175% after GPT-5.1.

    原文第3句

    ⬇︎ 下载 PNG𝕏 分享到 X
  • They turned a product bug into a piece of content. That's the difference between treating comms as damage control and treating it as something people actually want to read.

    原文倒数第2段

    ⬇︎ 下载 PNG𝕏 分享到 X
#OpenAI#AI communication#product comms#LLM bug
打开原文

For context, ChatGPT had a bug where it kept talking about goblins.

The "nerdy" personality mode was only 2.5% of all responses but caused 66.7% of the goblin mentions. Usage of the word "goblin" jumped 175% after GPT-5.1.

It was a real product" / X

This is really good comms. For context, ChatGPT had a bug where it kept talking about goblins. The "nerdy" personality mode was only 2.5% of all responses but caused 66.7% of the goblin mentions. Usage of the word "goblin" jumped 175% after GPT-5.1. It was a real product issue. Most companies would quietly patch this and put out a formal statement. OpenAI wrote a full blog post called "Where the goblins came from," broke down exactly what happened, and then posted it with an image that says "Never talk about goblins." Sam even tweeted "artificial goblin intelligence achieved." A play on AGI. People found this v funny. They turned a product bug into a piece of content. That's the difference between treating comms as damage control and treating it as something people actually want to read. Very few companies do this well. OpenAI, Anthropic, Stripe. I think more companies should study how OpenAI handles stuff like this. Not the AI part. The communication part.

Quote

Image 1: Square profile picture

OpenAI

@OpenAI

Apr 30

We’re talking about Goblins. openai.com/index/where-th

AI 可能会生成不准确的信息,请核实重要内容

This is really good comms. For context, ChatGPT had a bug where it kept talking about goblins. The... | Aadit Sheth(@aaditsh) | traeai