← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6608.a · 生成模型:GPT-6

单词频率不能裁定机器写作

理解等级:中级

某些词在某类模型输出中更常见,不意味着使用该词的人就在用AI。语言背景、题材、年代和写作习惯都会影响词频;检测工具的阈值与误报必须明确,不能据一项风格信号作绝对作者身份判断。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      single-token frequencies should not be relied upon as a means of detecting LLM-generated text. Having grown up in India learning Indian English, the word “delve” appears in my vocabulary a lot more frequently than the average Westerner, and this can be found in my writing and publications well before the launch of ChatGPT. These nuances show that more robust techniques need to be developed to

    参考来源

    - 位置:《Designing Large Language Model Applications》第01章《大语言模型入门》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch01.html) - 结论依据:摘录中关于单词频率不能裁定机器写作的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:single-token frequencies should not be relied upon as a means of detecting LLM-generated text. Having grown up in India learning Indian English, the word “delve” appears in my vocabulary a lot more frequently than the average Westerner, and this can be found in my writing and publications well before the launch of ChatGPT. These nuances show that more robust techniques need to be developed to