After 10 years of science communication since graduating from Tsinghua, how do I push back against AI?

Editor's Note

"Perhaps writing in the final era of human creation is also quite romantic and quite cool."

Liu Aiqi, known online as "laq is a hamster," has more than one million followers across all platforms, and his audio program "Science in Plain Words" has accumulated over five million plays.

Liu Aiqi was interviewed by China Science Daily. He said he has not fled from technology, but only reminds himself that he needs to hold fast to trust in people, loyalty to expression, and the warmth of face-to-face communication.

What follows is his own account. May every creator hear their own voice amid the flood of algorithms.

Becoming a Writer

In 2016, when AlphaGo burst onto the scene and swept the world of Go, I was preparing to leave Tsinghua University.

There is one scene I have always remembered: at the graduation dinner, a few friends and I sat around, and all we talked about was that program that had beaten every top human player. There was a constant anxiety in the lab—failed experiments were the norm, and being criticized by the "boss" was also the norm. That day I realized that procedural experimental research might also be the kind of work that AI would replace.

So I told myself that I had to choose a job that could not easily be quantified. I chose writing.

I have loved writing stories since I was a child. In elementary school, I once pretended to be sick to get leave, went home, and read Lu Xun. At the time I could not understand much of it; I just felt that people could still write essays with such verve—how cool!

I still remember reading Lu Xun's "The Lost Good Hell" when I was 8. The devil in the story says: "Friend, you are suspecting me. Yes, you are human! I will go look for wild beasts and evil ghosts..."

In high school I served as president of the literature club. My "chuunibyou" flared up, and I became wildly obsessed with absurdism—"The Bald Soprano," "Waiting for Godot," and the like. At that time I also wrote one whimsical little story every day, and only the girl who sat next to me liked reading them.

In 2012, with a score of 696, I was admitted to the School of Life Sciences at Tsinghua University to study life science and technology. In my freshman year I joined a remarkable structural biology lab, and after that I spent almost my entire undergraduate years in the lab. I arrived at the lab at 7:30 every morning and returned to the dormitory at 11:30 at night. If it was time to collect data, staying up all night was normal.

The days in the lab were full of anxiety, and that pressure needed an outlet. So I often went from the lab to a small café in the medical school, finished a little story, and posted it to Tieba. At the time I loved "science weird tales," and almost all my material came from the lab. I gradually discovered that quite a lot of people on Tieba were reading my stories. That also gave me the confidence to pursue writing after graduation.

At the time, outside the Tsinghua gate there was a crowded "urban village" called Shuimo Community. Because the rent was cheap, it was very crowded, and most of the residents were technicians, students doing their graduation projects, and postdocs. A friend and I shared a place for 1,200 yuan a month.

Fortunately, by then I had gradually become somewhat known on Zhihu, and those "science weird tales" brought me advertising opportunities. The monthly income was enough to support myself.

Back and Forth

In the second half of 2022, at an art competition at the Colorado State Fair in the United States, a painting generated with Midjourney (an AI image-generation tool) won first place in the digital arts/digitally manipulated photography category. That result directly angered a host of artists online.

My view at the time was actually optimistic about technological development but pessimistic about individual creation.

After ChatGPT appeared, I suddenly felt empty inside. Creation could let me prove my sense of existence; if writing, too, were replaced by AI, my sense of value would completely disappear. At that time, what exactly was my creation for?

Like Don Quixote, I charged at windmills with my lance, but the windmills did not care.

So I joined Science Popularization China. After the pandemic ended, I was responsible for the content of the science rumor-refutation section, later did account operations and event planning for a period, and then led the recording of courses for the Science Popularization China Creation Academy, teaching people how to do science communication. I did that for two and a half years.

Gradually, I saw another face of AI. With the emergence of AI, rumors actually increased. Family group chats were full of "how AI says..." which was in fact long-winded nonsense, and AI is extremely sycophantic. It rarely points out your mistakes directly; even when it thinks something is wrong, it still tactfully reminds you that "there is some truth to it," and then finds evidence to echo your views. Interestingly, the various AI "health rumors" my mother-in-law sent to the family group became the real-world source for my rumor-refutation work.

And AI's contribution to science communication itself is very limited. Finding materials and writing articles looks useful, but in reality there are a great many hallucinations—the more specialized the scientific field, the higher the probability that AI fabricates. Therefore, using AI to mass-generate content under the banner of science communication has become a very dangerous thing.

Some people may ask: why go back and forth?

Belief in "People"

The answer is "yes." In 2025 there was a study done by a team at the University of Zurich in Switzerland. They put 34 AI bots on the American Reddit forum, and they did only one thing: induce users on the forum to change their views. As everyone knows, people online are the hardest to persuade, but the AI succeeded and persuaded many people to change their positions. In the final tally, it was found that AI's efficiency in persuading people was 3 to 6 times that of humans persuading humans.

A paper in August this year presented a very interesting experiment. Scientists recruited 1,682 adults and had them do just one thing—read short stories. There were six stories in total: three written by humans and three written by ChatGPT. After reading, the human subjects rated them for creative quality and immersion. The result was piercing: the stories written by AI scored higher than those by humans.

But there was one exception in the experiment: as long as the subjects were told "this was written by a human," their ratings went up. What does that mean? Although we actually prefer reading stories written by AI, in our hearts we trust humans more, and we trust more that the person on the other side is a human creating. This is also why netizens get so angry once they discover that some game text was written by AI, or that some painting was painted by AI. It's not that the thing is bad—it's that you betrayed their trust.

The answer is "no." The above study also ran two other rounds of experiments, recruiting 905 adults to read two novels with similar themes, one written by a human and one written by AI, and then guess which one was written by a human. The first round's accuracy rate was 40%, and the second round's was 52%. In other words, humans' ability to tell the difference is less than or equal to a coin flip.

Attributing creation to oneself

Once I figured it out, I actually felt: perhaps doing writing in humanity's final era of creation is also quite romantic and quite cool.

So I reconsidered a question: what exactly is literariness?

In my freshman year, I took an elective course on understanding literature. At that time I believed one view: the core of literariness lies in moving people. You can say it touches the soul, or you can say it touches nerve cells—in any case, it has to touch me.

So, in the era of AI creation, if I read an article created by AI and am moved by it, does it have literariness? I think it does; at least at the reader level, literariness exists. So literariness is shifting from being jointly constructed by "author-reader" to being handed back to the reader, handed back to the person who experiences it.

And as for creators, which one of them isn't someone bursting with the desire to express? Authors not only long to speak but also long to be read. If they suddenly lost their readers, they would certainly feel forlorn. But in the end they will adapt, as long as their desire to express has not completely disappeared.

Han Yu said, "When there is injustice, one speaks out." From a literary perspective, I have something to say to this world, so I say it. When my expression aligns with the voice of the public, everyone will see me; when my expression aligns with the voice of the era, I will be remembered by the era. This desire to create is especially important in the AI era.

That being the case, what can I do as an author? My answer is to attribute creation to oneself. AI's literariness has already been handed over to readers; can I write the words I myself want to write, send out my own voice, and complete my own emotional journey? Attributing all perspectives to oneself can, on the contrary, help avoid being eroded by AI.

There is another direction I will stick to, and that is offline. In the AI era, humans still long for face-to-face communication; this is something evolution has carved into our bodies. For a period in the past, authors' offline exchange activities were once "disdained" because traffic was limited; but I happen to think that offline is a pure land that AI finds hard to influence.

Give literariness back to readers, give expression back to oneself, give people back to face-to-face. This is probably my resistance.

The 2025 survey data from "The Characteristics, Risks, and Countermeasures of Generative Artificial Intelligence Use by Primary and Secondary School Students—Based on a Survey and Analysis of 8,563 Primary and Secondary School Students in Seven Provinces (Municipalities) Nationwide" show that about 63.7% of primary school students in urban areas are using AI, while the proportion in rural areas is 60.2%. AI's impact on this era may be deeper than we imagine. The new generation of children has already become accustomed to using AI.

According to my observation, many parents treat AI as a positive tool and encourage their children to use it. One reason behind this may be that parents project their own anxiety about AI onto their children: AI has impacted their lives and their era, so they very naturally pass this unease on to their children, fearing that their children will fall behind.

In fact, to use AI better, people need to develop many logic skills and abilities in order to master it. But if, at the age when such skills should be developed, AI completes this process for the child, it will be hard for the child to truly possess these skills. Once the skills are missing, in the end they will not be able to use AI well either. Fearing that children will lose at the starting line actually makes it easier for them to lose at the starting line—that is the logic.

So, don't have too much AI anxiety. If you believe AI is definitely the direction of the future, then we will eventually get used to living with it. And by then, AI will definitely belong to most people, and its learning cost and usage cost will not be too high.

(Interview compiled by Hu Minqi, reporter for this newspaper)

1; OpenAI adjusts its security issue disclosure mechanism

2; The debate over whether to hit the brakes on AI is raging—are the tech giants serious?

3; When the mathematical fortress is conquered, will other disciplines usher in a moment of AI breakthrough?

1; Paper topic inspiration: "Sustainable Development: Agricultural and Forestry Waste Treatment and Reuse" research direction | MDPI Sustainability

2; The application of artificial intelligence in cultural heritage tourism: innovation, accessibility, and sustainability in the digital age | MDPI Heritage

3; LAF From "intelligent teaching assistant" to "disciplinary engine": how AI is reshaping the university classroom

4; ENG.Energy Professor Zheng Yun and Academician Zhang Jiujun of Fuzhou University, and Professor Yu Bo of Tsinghua University: Research progress on post-combustion carbon capture and conversion using advanced zinc-based metal-organic framework (Zn-MOFs) materials

5; "Addictive design" in the AI era: who is quietly stealing teenagers' attention? | MDPI Youth

6; Paper topic inspiration: "glioma" research direction | MDPI Brain Sciences

7; Soil moisture thresholds affect soil carbon dioxide emissions: a machine learning-based short-term soil carbon dioxide emission prediction method for climate-smart farmland | MDPI Sustainability

8; "Maize Sciences" | The research team of Researcher Fan Xingming of the Yunnan Academy of Agricultural Sciences reviews research progress on the genetic basis of maize resistance to southern leaf blight and the breeding application of molecular mechanisms of disease resistance

1; The western cryosphere is accelerating its "degradation": glacier melting sounds the alarm for alpine safety

2; The 6th National Agricultural Science and Technology Development Strategy Think Tank Alliance Conference was held in Lanzhou

3; The 2026 Nobel Prize is about to be announced; authoritative experts provide you with a comprehensive interpretation!

8; He became the first person in the national medical field to earn a doctorate without a paper

Famous Doctoral Dissertations 163: Nobel Laureate Dr. Peter Doherty's 1970 doctoral dissertation, University of Edinburgh

My elite-school complex and my elite-school experience

More>>