In debates about AI, there’s always a reminder that we must protect the ‘human factor’ from being overtaken by AI.

In our class on “AI and Journalism” at BNBU, we turned this idea on its head — because it’s precisely the ‘human factor’ that has contributed to the distorted view of China in Western media.

When we did a comparative analysis of human responses vs. LLMs, we arrived at some unsurprising insights: Humans reveal bias and emotions, while AI reflects discourse bias. So far, so good.

Western LLMs are far less critical of China than Western mass media

But after comparing responses from Western and Chinese LLMs, we found something interesting: both speak about the other sphere in a much more neutral, ‘objective,’ and less judgmental way than humans. Sure, LLMs still carry discursive system biases — the “human factor” is baked in, because, after all, humans built them! But thanks to the way we built them, they avoid extreme moral judgments, each subtly highlighting different angles: Chinese LLMs take a systemic or political lens (impact, divisiveness), while Western LLMs focus on image and personality (business, boldness). And what is particularly striking: Western LLMs are far less critical of China than Western mass media (or Western ‘experts’. They don’t get bogged down in moral or ideological takes — it’s all pretty descriptive and analytical. (Granted, I’m saying that with a pinch of salt — small sample, big claims.)

HAN Zilin is a student of Media and Commnunication Studies at Beijing Normal-Hong Kong Baptist University and the Editor-in-Chief of BNBU’s student magazine Converse.

These differences in emphasis — again, Western LLMs tend to focus more on image and personal traits, while Chinese LLMs highlight systemic or political contexts; Western models often approach topics from an economic or market-oriented perspective, while Chinese models lean towards political and societal structures — reflect the cultural and political biases in the training data. So LLMs don’t rise above their own cultural blind spots — they carry them along. True objectivity, after all, is a myth. But at their best, LLMs might help strip away some of the emotional charge and polarisation, nudging the focus back toward reasoned argument and factual exchange. That restraint reflects their programmed neutrality — an effort to appear as objective sources of information.

Our thought was simple: LLMs might benefit the conflict-hungry Western media when it comes to reporting on China. They could help de-escalate debates by filtering out ‘ideological noise’ and refocusing on a more neutral, analytical perspective. Such a ‘de-ideologization’ could create space for more rational, solution-oriented news articles and discussions — especially in today’s climate. A kind of new Enlightenment, perhaps?

In science, unlike in philosophy, nothing stands without evidence or empirical verification. So we put this to the test — using a recently published German text that both praises China’s development and cautions against it. Here it is (in English translation):

China is now a scientific superpower. The West should view cooperation as an opportunity, but remain cautious.

By Sven Titz

Modern research in China often comes dressed in striking architecture. In Hangzhou, a new centrifuge boasting record-breaking capacity now whirls material samples weighing up to 20 tons at speeds that subject them to 300 times the force of gravity. The machine will be used for materials research and earthquake studies, among other applications.

This centrifuge is just one example of China’s rapid emergence as a scientific superpower. According to a Nature-indexed ranking of research publications, eight of the ten most productive scientific institutions worldwide are now in China, with Harvard University and Germany’s Max Planck Society rounding out the top ten. Chinese research quality has surged, particularly in engineering, electronics, materials science, physics, and chemistry. While the life sciences still lag behind, progress is accelerating.

Remarkably, this scientific ascent is unfolding under an authoritarian government, posing challenges for liberal democracies regarding collaboration, intellectual property, and potential espionage. Doctoral Growth Fuels Success China’s growth is fueled in part by education. For roughly two decades, Chinese universities have produced more PhDs in STEM fields — science, technology, engineering, and mathematics — than the United States. In 2025, Chinese institutions are expected to graduate over 77,000 STEM doctorates, compared to around 40,000 in the US.

Quantity alone does not tell the full story. Historically focused on applied research, China now invests heavily in high-quality, well-funded basic research. The record centrifuge, and upcoming projects such as the deep-sea research vessel Meng Xiang, demonstrate China’s commitment to foundational science. The vessel will probe continental plate movement, past climates, and deep-sea microorganisms, with a long-term goal of drilling through the Earth’s crust to reach the mantle — a feat unmatched by current international competitors. Support for Fundamental Research Chinese leaders emphasize nurturing young scientists for long-term basic research. Dou Xiankang of the National Natural Science Foundation stresses creating environments that foster independent innovation. Huawei founder Ren Zhengfei echoes this sentiment, noting that breakthroughs require a strong foundation in basic research.

China’s historical and cultural focus on education also contributes. STEM disciplines are highly valued in families, giving Chinese students a long-standing advantage over many Western peers, where interest in rigorous scientific fields has been declining. Collaboration vs. Risk For democracies, collaboration with Chinese institutions presents a dilemma: the potential for intellectual theft and military applications is real, yet disengagement risks missing out on major scientific advances. Cooperative research in non-sensitive areas, such as deep-earth exploration or climate-resilient crops, can yield mutual benefits without triggering security alarms. Still, political tensions — over Taiwan, Ukraine, and trade, among other issues — mean that scientific cooperation with China will remain a delicate balancing act. Western countries may often prefer alternative academic partners in the region, such as South Korea, Japan, or Taiwan, where research excellence exists without comparable political risks.

China’s rise as a scientific powerhouse is undeniable. With record-breaking technology, a growing base of STEM-trained PhDs, and strong institutional support for basic research, the country is reshaping the global scientific landscape. Yet for liberal democracies, engaging with this new powerhouse requires caution, strategy, and careful assessment of both opportunities and risks.

We used ChatGPT to sift through the ideological biases. Here’s what it came up with:

• “Espionage and ideas theft” as a one-sided narrative: The text repeatedly warns about China being a source of “idea theft” or “espionage,” framing it as inherently dangerous. This is a bit of a cliché and selective: historically, Western countries – including the U.S. and Europe – have also engaged in extensive scientific and industrial intelligence operations. Treating China as uniquely at fault reflects a geopolitical bias rather than an objective assessment.

• Western moral superiority assumption: The text implicitly suggests that Western liberal democracies are morally and scientifically superior. (German philosopher Hans-Georg Moeller btw, in an excellent video, argues that this assumption, what he calls ‘the superiority complex,’ is the main reason why the media and experts have been so successful in shaping public opinion in Germany and neighboring German-speaking countries with their “non-narrative” narratives – like “China is bad, Russia is even worse.”) This superiority assumption overlooks issues within Western research itself: corporate influence, military-funded projects, and even fraudulent publications. China is portrayed as an exception, rather than part of a global system with similar pressures.

• Overstated political risk: Phrases like “egg dance for the liberal West” exaggerate the dangers of scientific cooperation, implying catastrophic consequences for any engagement. In reality, international collaboration is already managed with safeguards, IP agreements, and selective partnerships. The risk is inflated while the potential benefits are downplayed.

• Assuming science exists only within a liberal framework: The critique treats political freedom as a prerequisite for valuable science. However, empirical evidence shows that China’s research output and breakthroughs are globally competitive, suggesting that highly regulated or state-directed systems can also produce top-tier science.

• Contradiction in praising achievements while warning against them: The text celebrates China’s scientific advances (centrifuges, deep-sea drilling, STEM graduates) but then frames them as threatening. This inconsistency weakens the analytical credibility—it reads more like a cautionary opinion piece than a balanced factual account.

• Selective comparison: Western weaknesses (declining STEM engagement, visa issues, bureaucracy) are framed as problems that benefit China strategically, yet the “threat” narrative dominates, creating a skewed perspective.

Then we asked ChatGPT to filter out those biases. The result, rather amusingly, ended up sounding like straight-up Chinese propaganda:

China’s Rise as a Scientific Power: From Record-Breaking Centrifuges to Global Ambitions

By Stan Sits

Modern research in China is often characterized by striking, innovative architecture. In Hangzhou, a new centrifuge with record-breaking capacity spins material samples weighing up to 20 tons at speeds that subject them to 300 times the force of gravity. This machine will be used for materials research, earthquake studies, and other applications.

This centrifuge is just one example of China’s rapid emergence as a scientific powerhouse. According to a Nature-indexed ranking of research publications, eight of the ten most productive scientific institutions worldwide are now based in China, with Harvard University and Germany’s Max Planck Society rounding out the top ten. Chinese research quality has surged, particularly in engineering, electronics, materials science, physics, and chemistry. While the life sciences still lag behind, progress is accelerating.

This scientific growth is being fueled in part by China’s commitment to education. For nearly two decades, Chinese universities have produced more PhDs in STEM fields – science, technology, engineering, and mathematics – than the United States. In 2025, Chinese institutions are expected to graduate over 77,000 STEM doctorates, compared to around 40,000 in the U.S.

But quantity alone does not tell the full story. Historically focused on applied research, China has increasingly invested in high-quality, well-funded basic research. The record-setting centrifuge and upcoming projects like the deep-sea research vessel Meng Xiang showcase China’s dedication to foundational science. This vessel will study continental plate movements, past climates, and deep-sea microorganisms, with a long-term goal of drilling through the Earth’s crust to reach the mantle – an achievement that remains unmatched by current international competitors.

Support for fundamental research is strong. Chinese leaders emphasize the importance of nurturing young scientists and fostering environments that encourage independent innovation. Dou Xiankang of the National Natural Science Foundation has stressed the need to create spaces that cultivate creativity. Huawei founder Ren Zhengfei echoes this, highlighting the necessity of a robust foundation in basic research for achieving breakthroughs.

China’s cultural emphasis on education also plays a role. STEM disciplines are highly valued, giving Chinese students a long-standing advantage over many Western peers, where interest in rigorous scientific fields has been declining.

China’s rise as a scientific leader is undeniable. With cutting-edge technology, a growing base of STEM-trained PhDs, and strong institutional support for fundamental research, the country is reshaping the global scientific landscape.

Why is that? How come a text scrubbed clean of bias — a supposedly ‘purified’ version — suddenly reads like propaganda? Is Sven Titz secretly an admirer of China, just unable to say so openly in the West for fear of being cancelled? Maybe his praise was simply … disguised?

We can’t look inside the author’s head. It’s entirely possible that the fear of backlash or career damage shaped his writing — after all, in the West, free speech is increasingly more rhetorical than real. It’s no coincidence that, as the Allensbach Institute found, fewer and fewer people in Germany feel they can really speak their minds anymore. Or, as former China-born journalist  at Deutsche Welle puts it in her book: “Only the Right Opinion Is Free.” But whether that influenced Titz while writing — that’s not empirically verifiable, so it’s not science.

Secondly, as we’ve already hinted, such “cleaning” isn’t really possible. Remove the perspective, and there’s nothing left. LLMs, just like us, are mirrors of their training. Yes, the fact that they sound less polemical and more analytical than many humans — less partisan, less emotional—- is genuinely valuable, especially in heated times like these. But that shouldn’t blind us to the inherent necessity of perspective, or more dramatically: bias.

Large-scale science is a form of soft power

And then there’s the question of large-scale science, which the text itself deals with. It’s always also a form of soft power. Centrifuges, deep-sea vessels, “massive PhD systems” — all of these express who sets global standards (according to Titz: China) and who, ultimately, owns the future (again China, it seems).

In the ChatGPT rewrite, a kind of technocratic neoliberalism came through — the idea that technical progress is neutral and inherently good. A classic Silicon Valley philosophy, unsurprisingly, given ChatGPT’s origins. Yet this “technocratic liberalism” isn’t just a Western thing. After all, who’s building AIs — LLMs — in China? Next, we asked DeepSeek and Kimi, two Chinese LLMs, to produce their own rewrites of the China-critical article and compared the results. Both came up with strikingly different versions.

DeepSeek, once it stripped away the Western lens, promoted harmony, openness — and a similar progress narrative. We called the underlying ideology “developmental nationalism.” The tone was dry, formal, bureaucratic — very “state-style.” Not exactly thrilling reading, but anyone aspiring to draft government publications would find their perfect tool here.

Kimi, on the other hand, was a delight. Its version was both analytical and provocative — genuinely “readable,” as they say in the West. Why? Because Kimi didn’t even try to erase ideology altogether. Instead, it acknowledged the tension, framing it not so much as ideology but as a systemic contrast: China’s planning versus Western hesitation. (A contrast that the recent book “Breakneck” by Dan Wang describes as one between the “engineering state” — China  — and the “lawyerly state” — the USA.) This political framing reads as analytical rather than partisan. It gives readers something to think about, rather than something to believe in.

What really matters is transparency

And with that, poor Sven Titz was — well, partly — rehabilitated. What really matters is transparency: writing from a clearly stated position. In his case, for example: “This piece views China as a risk because, frankly, admitting otherwise could jeopardise my career and social standing.” Rather than pretending there’s no position at all. And I, too, felt oddly relieved. In my own ‘human’ version of the text, I had simply flipped Titz’s argument — turning his seemingly harmless assessment on its head and reworking the claim that political tensions make scientific cooperation with China a “delicate balancing act”:

While Western democracies often focus on geopolitical risks or intellectual property concerns, these constraints can limit engagement with emerging scientific opportunities. China’s rise challenges the assumption that global leadership in science must align with U.S.-led rules or governance models. Instead, it demonstrates that innovation can thrive under different political systems, offering avenues for the world to benefit from new discoveries.

Certainly, from a Chinese perspective, cooperation with the West might seem like a delicate balancing act. Political tensions – fueled by the West’s ongoing disregard for international law, including Israel’s supported genocide, sanctions against nations striving for self-determination, attempts to instigate regime change in Iran, as well as the destabilization of sovereign governments in Latin America and Africa – make it clear that the West’s own actions often complicate global cooperation.

In the West, this kind of relativising — putting things in relation — is often denounced as ‘whataboutism:’ a rhetorical move where one responds to criticism by pointing to another, often similar, wrongdoing. True, it is questionable when it merely changes the subject instead of engaging with the criticism. Or when it slips into moral relativism (“everyone’s guilty”), or serves as a strategic tool to silence critique rather than deepen understanding. I call that “lazy whataboutism.” But in my view, it’s legitimate when it’s not just reactive but analytical — as in this case, where it mirrors the rhetoric to expose bias and reveals a double standard: showing that the moral logic applied to China isn’t applied to the West; and when it adds context — demonstrating that the same structural forces operate on both sides — and challenges hidden assumptions, such as the idea that only China introduces “political risk,” while Western interference somehow isn’t political. And of course, this rhetorical inversion also makes for better reading. Because it makes clear where one is speaking from and doesn’t pretend to be ‘objective’ or neutral. 

So again, does that redeem Sven Titz’s text, the one we set out to “clean”? Not exactly. His warning about China still comes off as reflex rather than reflection — a kind of rhetorical insurance policy. After all that praise, he seems to feel obliged to add a pinch of alarm and some easy-to-digest stereotypes, just to stay on the safe side. Still, it does make one a little more forgiving.

Maybe AI really can give us a hand — I know, it doesn’t literally have one — at a moment that feels this tense. Think of it as an ongoing reality check, gently nudging us to stay focused on meaningful dialogue and to keep trying to see things from each other’s perspectives. To paraphrase two AI experts we discussed in class, Katharina Schell (APA) and Victoria Nash (Oxford Internet Institute), journalism needs more balanced reporting on China — less emphasis on emotional extremes, more on how China actually works (and the real-world consequences of its actions). Or, to borrow Titz’s own vocabulary, these ‘AI guardians’ can highlight the risks of oversimplification and cliché—the danger of drifting into distractions or easy narratives. And no, that doesn’t mean problematic issues shouldn’t be raised. If anything, it’s the journalistic version of what the West so neatly calls “de-risking.”

If we actually let it, that is.

Image created by Kathy and https://muryou-aigazou.com/