请问HN：有没有人能解释一下为什么OpenAI的推理现在会显示非英语的思维？
人们已经注意到一段时间，谷歌的Bard/Gemini经常插入随机的印地语/孟加拉语单词。[0]<p>我刚刚在一个o3-pro的思考过程中发现了这一点：“并为低难度进行定制。 কাজ করছে！”<p>最后一组字符显然是孟加拉语，意思是“正在工作！”<p>我只是觉得很奇怪，类似的“错误”出现在多个不同的模型中……有没有人知道这些其他语言是如何渗入的，训练方法或推理是什么？<p>[0] https://www.reddit.com/r/Bard/comments/18zk2tb/bard_speaking_random_languages/
查看原文
People have noticed for a while now that Google's Bard/Gemini has inserted random hindi/bengali words often. [0]<p>I just caught this in an o3-pro thought process: "and customizing for low difficulty. কাজ করছে!"<p>That last set of chars is apparently Bengali for "working!".<p>I just find it curious that similar "errors" are appearing from multiple different models... what is the training method or reasoning that these alternate languages can creep in, does anyone know?<p>[0] https://www.reddit.com/r/Bard/comments/18zk2tb/bard_speaking_random_languages/