OpenAI's AI reasoning model 'thinks' in Chinese sometimes, no one knows why

It's not unusual for someone who's fluent in multiple languages to have internal thought processes that are associated with one language or another and to mix them whether consciously or unconsciously.

> Other experts don’t buy the o1 Chinese data labeling hypothesis, however. They point out that o1 is just as likely to switch to Hindi, Thai, or a language other than Chinese while teasing out a solution.

It could be showing an abstraction over languages and choosing the right words/language is something that could come later in the process rather than all the way through. If the 'chain of thought' (to use another anthropomorphization) be from another language. It's not proof but it is interesting.