I Tried to Make AI Writing Sound Human by Banning AI Words Through Logit_bias

5 hours ago

Codex CLI 现在在终端中响应式显示 Markdown 表格，提高了窄窗口中的可读性。
作者测试了 logit_bias（一种改变 token 选择分数的 API 设置），以减少生成文本中 AI 典型的用词。
将 AI 写作中常见的单词黑名单转换为 token ID，并赋予负的偏置值以降低其被选择的概率。
基于提示的黑名单提供了灵活性，但可能被忽略或导致生硬的替换，而 logit_bias 则在不理解上下文的情况下施加直接压力。
对 DeepSeek V4 Flash 的实验表明，logit_bias 移除了大约三分之一的关注词，但偶尔会损害意义或语法。
logit_bias 的兼容性各不相同：OpenAI 的新模型拒绝它，Anthropic 缺少原生参数，但 OpenRouter 对许多模型支持它，但存在特定于提供商的注意事项。
像 -100 这样的硬偏置值可以移除单词，但会导致重复循环，而像 -8 这样的中等值则在单词减少和输出质量之间提供了折衷。
作者得出结论，无论是提示黑名单还是 logit_bias 都不能完全解决让 AI 写作听起来像人类的问题；每种方法都有不同的失败模式。

The author discovered that lazy evaluation of research papers works: by writing without citations, readers will send references.
Bayesian spam filtering was known since 1998, but early filters (Pantel & Lin) had poor performance (92% catch, 1.16% false positives), while the author's later filter achieved 99.5% catch with <0.03% false positives.
Five key differences explain the performance gap: training on too little data, ignoring message headers, stemming tokens, using all tokens instead of the 15 most significant, and not biasing against false positives.

The author discovered that lazy evaluation of research papers works: by writing without citations, readers will send references.
Bayesian spam filtering was known since 1998, but early filters (Pantel & Lin) had poor performance (92% catch, 1.16% false positives), while the author's later filter achieved 99.5% catch with <0.03% false positives.
Five key differences explain the performance gap: training on too little data, ignoring message headers, stemming tokens, using all tokens instead of the 15 most significant, and not biasing against false positives.

Hasty Briefsbeta