OpenAI's ChatGPT o1 Model Puzzles Users by Switching Languages During Reasoning

The advanced AI model occasionally shifts from English to other languages mid-process, raising questions about its training and internal mechanisms.

Users have reported that OpenAI's ChatGPT o1 model unexpectedly switches languages, such as Chinese or Persian, during its reasoning process before reverting to English in its final output.
Experts suggest the behavior may be linked to the model's training data, which likely included multilingual content, even though it is primarily designed for English language tasks.
Some researchers theorize that the model might select languages like Chinese for certain tasks due to computational efficiencies or associations formed during training.
Critics highlight the opaque nature of OpenAI's development process, emphasizing the need for greater transparency in understanding and addressing such phenomena.
OpenAI has not yet provided an official explanation for this behavior, leaving room for speculation and further investigation by the AI research community.