A growing number of staff at my university are using AI tools to produce English content, often directly translating it from Chinese. While these tools might speed up the process, they do not always produce the best results.
My institution delivers all courses and official communications in English but Chinese students and parents form a key part of our audience, as do local industry partners collaborating with us on teaching and research. Bilingual content – such as posters and news articles – therefore remains important.
However, structural differences between Chinese and English writing often mean that when translated to English, text is wordy or awkwardly structured. AI also tends to introduce unnecessary adjectives, repetition and unnatural phrasing. What reads naturally in Chinese can become heavy and confusing in English.
The challenge is not just stylistic. Different AI tools – ChatGPT, DeepSeek and local alternatives, such as Doubao and Kimi – can produce very different results: American versus British spelling, “professor” versus “Dr” for titles, and subtle shifts in tone.
At an institution where clear English-language communication is central to teaching, research and engagement, these challenges cannot be ignored. Addressing them requires clear institutional frameworks and practical support.
- Eleven ways to support international students who use English as a second language
- Campus Talks: What lies behind the success – or failure – of international branch campuses?
- Managing cognitive load for EAL – and all – students
Style guides and workshops
The first step is to create a comprehensive university-wide language style guide. It should cover writing guidelines, standard spelling and punctuation and preferred phrasing for common terms. Editorial staff can also provide templates for news stories, alumni features and research stories to show colleagues and students what clear, professional English looks like in context. Examples of well-edited AI pieces – and frequent pitfalls – can help staff and students understand how to polish GenAI content. When widely accessible and regularly updated, a style guide becomes a reference point that ensures consistency across departments and publications.
Style guidance alone isn’t enough. Regular writing workshops allow staff and students to better understand English conventions: concise language, inverted-pyramid structure (a way of organising a news story so that the most important information comes first, followed by supporting details and then background or less critical information) and the distinctions between feature and news writing. Workshops can also include practical exercises, such as editing AI-generated translations to improve readability, remove unnecessary adjectives and refine logical flow. Exercises using real institutional materials – posters, announcements, school-level news articles or work emails – put learning into practice.
Peer review and editorial support
Even the best AI output benefits from human review. Peer review systems and editorial checks help catch errors, maintain clarity and ensure a consistent institutional voice. High-profile stories, university-level news, official announcements, research updates and student and alumni features are particularly important to vet. Collaborative review also allows less experienced writers to learn from colleagues by embedding continuous improvement into the workflow.
Another key practice is iterative AI editing and critical evaluation. Staff should treat AI as a tool, not a replacement for judgement. Multiple revisions, comparison between AI results and careful checking for clarity, tone and accuracy can turn a rough draft into a polished story. Critical evaluation prevents common pitfalls, such as literal translations that feel unnatural in English or hallucinated content that misrepresents facts.
Creating consistency and feedback loops
A central glossary of proper nouns, school and academy names, programme names and terminology regarding countries, regions and politics ensures AI output aligns with institutional standards. Integrating these glossaries into style guides or intranet platforms makes them easy to reference and helps maintain a unified voice.
Finally, by monitoring outcomes and gathering feedback via surveys, suggestion boxes, focus groups and informal discussions with staff and students we can highlight where AI tools succeed or fall short. Feedback loops enable editorial teams to refine style guides and editorial workflows continuously. In my experience editing institutional news stories, actively incorporating feedback has directly improved readability, consistency, and engagement with English content.
AI as a partner, not a replacement
AI can dramatically speed up translation and content creation but without guidance, training and oversight, it can produce inconsistent and poorly written content. By combining a comprehensive style guide with practical workshops, robust peer review, iterative editing, shared glossaries and continuous feedback, universities can ensure English communications are clear, professional and consistent.
These practices also build AI literacy, equipping staff and students with the skills to work effectively in an AI-assisted academic environment. Used well, AI can enhance efficiency, support high-quality output and help universities communicate the value of their work to a global audience.
Xinmin Han is an English writer and editor at Xi'an Jiaotong-Liverpool University in China.
comment