AI translation errors highlight need for higher education reform

Recognizing errors in AI-generated translations and providing precise prompts for correction now require higher educational standards, a political adviser said during the annual two sessions.

Jiang Yajun, a member of the National Committee of the Chinese People's Political Consultative Conference and former vice-president of Xi'an International Studies University in Shaanxi province shared his experience using artificial intelligence in translation work during a group discussion on Friday.

Jiang, an expert in translation and linguistics, has been focusing on cross-culture communication by translating academic works in fields such as Chinese culture, history and architecture.

He noted that when he first used AI for translations, he found that ChatGPT often misinterpreted Chinese content due to the differences in sentence structures between Chinese and English.

Generally, most of the materials AI deeply learns during its training period are in English.

"Chinese language relies on comma-linked sentences, which makes it difficult for AI to accurately grasp the meaning of the texts," Jiang said.

In addition, AI struggles to comprehend Chinese history and its unique divisions, like slave society and feudal society, which are not mirrored in Western historical frameworks, he said.

Effective cross-cultural communication is challenging, as it requires more than just translating into another language to ensure people around the world fully understand, he added.

In the information era, higher education should prioritize critical thinking, rather than merely teaching definitions and textbook knowledge in the classroom, according to the national political adviser.

Fang Biling contributed to the story.