A lot of meaning is derived from culture, so nuance and context are more of a struggle for AI, but the base pictograms are easier at the surface level, that is true.

Conversely, on the Chinese internet, people believe that due to the poor quality of Chinese corpora (spam and content censorship), using English prompts and requiring the model to respond in English would yield better output quality. Those who firmly believe this argument often also believe another ...

I assume you mean ‘Chinese characters’.