Seth Godin 谈人们和 AI 都不说出口的免责声明
Things people don’t say
- 1Seth Godin 指出营销者与政客因默认和恐惧，不说“我可能错了”“这可能出于私利”这类坦白
- 2他认为人们默认机器不欺骗不误导，AI 缺少这类声明而事后发现出错时会令人失望
- 3他建议在读广告、相信诈骗短信或与 LLM 交互前，先把这类免责声明读一遍
- 4他提出信任是重要文化的基石，商业与政治长期操纵信任，如今机器正在效仿人类
作者把 AI 不说“我可能错了”归因于人类沟通的老毛病，营销与政治早已习惯省略免责声明，机器只是效仿。
- 影响：对外内容的确定性口吻与 AI 生成内容叠加，出错时损耗的是品牌信任；把 LLM 输出当作可靠事实接入流程，风险会顺着“机器不说谎”的默认预期放大。
- 参考：可考虑盘点对外内容由 AI 生成的清单并试加不确定性说明；在用 LLM 辅助判断的环节，值得先定一份人工核验要点再采信输出。
- 依据：本文为 Seth Godin 个人观点，论断来自对营销沟通与 AI 交互的观察归纳，文中未引用数据或独立验证。
Marketers and politicians too.
We don’t say them because it’s assumed. And because we’re afraid.
“I might be wrong.”
“I’m just guessing”
“This might be selfishly motivated.”
“It could be that I’m hallucinating.”
“I heard this somewhere but I don’t know firsthand that it’s true.”
“I know I sound certain, but I’m not.”
When an AI doesn’t say these things, we’re frustrated when they turn out to be the case. Because our default in interactions with machines is that they don’t dissemble, confuse or mislead.
Perhaps we should print these caveats out and read them before we read an ad, believe a scammy text or interact with an LLM.
Trust is the building block of cultures that matter. We have long manipulated trust for commercial and political ends, and now machines are following our lead.
来源：Seth Godin · seths.blog