Meng Zhongjie, president of Shanghai International Studies University and a deputy to the Shanghai municipal people's congress, used his platform at this year's Two Sessions to press a simple but consequential point: the humanities cannot stand aside as artificial intelligence reshapes knowledge production. He argued that liberal arts disciplines should proactively prepare high-quality corpora, collaborate with AI systems to solve practical problems, and position themselves at the centre of a broader disciplinary transformation.
His remarks reflect an intensifying debate inside China about the role of the so-called "new humanities" (新文科): whether humanities scholars will be sidelined as AI automates routine interpretation, or whether they will seize the opportunity to re-skill, provide essential cultural and linguistic datasets, and influence the design of AI models. Meng framed this as an urgent, operational task — universities must curate the textual and multilingual resources that underpin large language models, and build institutional relationships with the technology sector.
Practical implications are immediate. Quality corpora are the raw material for Chinese-language and culturally informed models; who controls and curates that material will shape the biases, priorities and capabilities of future systems. Meng’s call to "actively cooperate" with AI developers signals a push by academia to be a partner in model training, rather than a passive source of content or a critic on the sidelines.
There are also political and professional stakes. The Chinese state has invested heavily in AI as a national priority, and education policy has promoted interdisciplinary programmes that blur the line between humanities and STEM. Meng’s appeal to attract young talent and "break down fences" between disciplines speaks to universities' need to redesign curricula, research incentives and career paths so that humanities graduates can work alongside engineers and data scientists.
Yet the proposal carries trade-offs. Deep engagement with AI firms can channel humanities research toward instrumentation and data-production tasks, potentially weakening critical, normative inquiry. At the same time, tighter control over corpus quality invites questions about gatekeeping, censorship and whose voices are preserved for training models. Balancing scholarly independence, cultural stewardship and strategic collaboration will be a central challenge for Chinese universities.
For international observers, Meng’s comments are a reminder that the geopolitics of AI is not only about chips and compute. Language, literature and cultural datasets are strategic assets. If Chinese universities systematically curate and supply high-quality corpora to homegrown models, Beijing-backed research and products will reflect distinct linguistic, normative and epistemic contours, complicating efforts to build globally neutral AI systems.