Urgent.News

What's breaking now, across thousands of outlets.

AI

AI can make China’s political language sound more threatening, scholar warns

China is being urged to use human translators to refine AI renderings of its political messaging, amid concern that literal translations can make Beijing’s language sound more threatening abroad. An article in a state-affiliated academic newspaper said the country should strengthen its translation capacity and guard against AI-generated distortions to improve how its policies and intentions were…

AI can make China’s political language sound more threatening, scholar warns

China's political messaging could become more threatening in foreign eyes due to literal translations made by AI, a scholar warns. A state-affiliated academic newspaper article suggested that the country should enhance its translation capacity and prevent AI-generated distortions to ensure its policies and intentions are understood overseas.

Professor Li Yan from Xian Fanyi University emphasized that AI should be complemented by human translators, as the value of translation extends beyond language itself. In the AI era, a nation's translation abilities have become a key indicator of its cultural soft power and capacity to engage in global governance, according to Li.

She called for Chinese translators to strengthen their skills in refining AI translations according to context, audience, and purpose. For instance, the Chinese idiom "taoguang yanghui," which historically conveyed Deng Xiaoping's foreign policy strategy of avoiding confrontation and prioritizing domestic stability, is often mistranslated in English as "hide your capabilities, bide your time."

This misleading interpretation could fuel perceptions that China aims to conceal its strategic ambitions until it is strong enough to challenge the United States. Li proposed recasting the idiom as "maintain humility while building strength" to better reflect China's diplomatic principles of peaceful development. The scholar also stressed the importance of recognizing and interpreting culturally specific symbols, metaphors, historical references, and political ideas in translations.

An example given was the 1980s reform slogan "mozhe shitou guohe," which could be better conveyed as "crossing the river by feeling the stones" to aid foreign comprehension of China's political-economic concept. Li's recommendations come as Chinese authorities and scholars place increasing importance on translating politically charged phrases and culturally specific concepts, which some view as part of a broader contest over international narratives and public opinion.

The challenges of translation have previously complicated high-stakes diplomacy, especially when languages and political cultures are vastly different. A 2021 Rand Corporation report found that several highly skilled China experts, who primarily rely on Chinese-language sources, inadvertently introduced a "hawkish" distortion in their translations and interpretations of Beijing's terminology.

Similarly, a political scientist at Helmut Schmidt University argued that Western governments have continued to misinterpret China despite having ample Chinese-speaking personnel and AI translation capabilities. Li warned that China must mitigate cultural misreadings, logical errors, and value-laden biases in AI translations, while actively embracing AI tools and developing the necessary skills to effectively work with them and manage associated risks.

Her article, published in Chinese Social Sciences Today, formed part of an ongoing research project funded by China's National Social Science Fund.

Written by urgent.news from Reuters Business via SCMP's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at scmp.com →

More in AI

The Gate Said No. Now What? A Triage Procedure for Rejected Agent Patches

A gate that rejects a patch is only half a policy. The other half is what happens after the rejection. In most pipelines, a failing agent patch produces one of three outcomes: a human stares at the…

  • Gate rejects patch, leading to three outcomes: human review, blind rebuild, or test deletion
  • Proposed triagegate.py script re-runs failing test, compares fixture hashes, freezes flakes
  • Class A: deterministic regression, Class B: fixture drift, Class C: flake with quarantine ledger

"My Agent Refused 96 Times": Building Self-Editing Agents with Hard Failure Modes

Originally published on tamiz.pro . In the early days of shipping LLM-based agents, we optimized for output volume. If the model could not find the answer, it often generated a plausible one anyway. This is the "yes-man" problem.

  • Agent refused 96 valid questions in testing
  • Demonstrated hard failure mode for insufficient context
  • Shifted focus to deterministic verification

More from Saturday 29 August →