HN 提问：你还能分辨出你母语中的 AI 生成文本吗？
1 分•作者: h_mirin•大约 1 个月前
我是日本人，我仍然通常能通过词语选择来辨别出人工智能生成的日语。有些词，比如“实务”和“账簿”，出现的频率远超正常。像“奏效”和“刺中”（引申为引起共鸣，切中要害）这样的口语词，在通常情况下会有其他词语作为首选，但却被用在了这里。它们相当于英语中的“delve”。
我不是英语母语者，但我的印象是，近期人工智能生成的英语有了很大进步，比以前更难辨别了。这和英语母语者的看法一致吗？
其他语言呢？中文一定是接受了大量训练的语言之一，所以我认为它应该和英语一样进步显著。那么西班牙语、印地语以及其他广泛使用的语言呢？母语者还能辨别出来吗？
另外，我认为一旦脱离英语，你使用的模型仍然至关重要。我第一次感受到“奇点”的时刻之一，就来自 Gmail 的 AI 草稿功能。它显然已经吸收了日本企业邮件文化的惯例，包括其中的官僚主义。
如今，Anthropic 的 Opus 4.6（在编码会话之外使用 Fable 5）能够写出相当不错的长篇日语，而 Google 的模型在更广泛的语域中表现也很好。OpenAI 的模型，Opus 4.8/5 和 Fable 在编码会话中听起来就太“极客”了。
很想听听你们的想法。
查看原文
I'm Japanese, and I can <i>still</i> usually spot AI-generated Japanese by its word choice. Some words, like 実務 (practical work) and 帳簿 (a ledger), turn up far more often than they should. Casual ones like 効く (to work well, to be effective) and 刺さる (to pierce > to resonate, to hit home) get used where another word would normally be the first choice. They're the Japanese equivalent of "delve."<p>I'm not a native English speaker, but my impression is that AI-generated English has improved a lot recently and is harder to pick out than it used to be. Does that match what English native speakers see?<p>What about other languages? Chinese must be one of the most heavily trained languages, so I'd expect it to have improved as much as English. And what about Spanish, Hindi, and other widely spoken languages — can speakers still tell?<p>Also, I think which model you use still matters far more once you leave English. One of my first singularity moments came from Gmail's AI drafting feature. It had clearly absorbed the conventions of Japanese corporate email culture, bureaucracy and all.<p>These days, Anthropic's Opus 4.6 (and Fable 5 outside coding sessions) writes reasonably good long-form Japanese, and Google's models hold up across a wider range of registers. OpenAI's models, Opus 4.8/5 and Fable in a coding session sound way too geeky.<p>Curious to hear your thoughts.