A few days ago I had dinner with an old classmate who teaches in a university's Chinese department. He vented to me about something, and after hearing it I was stunned for quite a while.
He said that grading undergraduate thesis papers now feels more and more like dealing with a machine. Not that the students write like machines, but that what they submit is obviously machine-written. The wording is flawless, the structure is neat and orderly, and the abstract, introduction, literature review, and conclusion all fit together perfectly. But you just can't sense any "human touch." There are no stumbling traces of thought, none of that authenticity of wandering off track while writing and then pulling yourself back, and not a single flash of insight that makes you think "this student really read the books."
He tried asking a student, how did you think through the logical transition in your second chapter? The student hemmed and hawed for a long time, then finally told the truth: Professor, I first entered the keywords, AI generated three versions of an outline, I picked the one that looked good to me, then had it expand each section, and I deleted the repetitive parts.
My old classmate said that at that moment he didn't know whether to be angry or sad.
This isn't actually new. From ChatGPT to various domestic large models, and then to AI tools specifically designed for academic writing, Chinese department students today have far more "cheats" at their disposal than we did back then. In the past, to write a paper you had to go to the library and flip through old periodicals, sift through literature one by one on CNKI, and force out the first sentence facing a blank Word document. That process of forcing it out was itself the process of thinking. You couldn't write it because you hadn't thought it through yet. Once you thought it through, the sentences would come naturally.
But now it's different. AI writing assistants can give you a text that "looks like you've thought it through" in three seconds. It makes the shell of thinking look beautiful, but inside it's empty. Students take this shell, make a few tweaks, and submit it. After submitting enough times, they start believing it themselves—believing they really know how to write papers.
That's where the problem lies. Writing a paper, in essence, is not about producing text—it's about training thought. When you're sorting through literature and discover that Zhang San and Li Si's views conflict, you have to figure out why they conflict, who makes more sense, and what your own position is. This process is painful, repetitive, and often involves overturning everything and starting over mid-writing. But it is precisely this pain that transforms a Chinese department student from "someone who can read" into "someone who can think."
AI skips this painful process entirely. It gives you a finished product, and you reverse-engineer it, thinking you've taken a shortcut. In reality, you've bypassed the only path that could help you grow.
Another friend of mine who works as an editor at a publishing house said something similar. He said that many of the submissions from young authors he receives now have an "AI flavor"—the word choices are precise but lack warmth; the logic is smooth but has no temper. People who come out of Chinese departments should be the group most sensitive to language. You know the subtle difference between this word and that word, you know where to break a sentence to make someone's heart tremble. These things, AI still can't give you. But if you go long without using your own linguistic intuition to judge and polish, your sensitivity will degrade. Just like people who rely on navigation for a long time—their sense of direction will definitely get worse.
Of course, I'm not saying AI writing assistants are monsters. Tools themselves are neither good nor bad. Using them to look up materials, organize thoughts, do translations, or even have them check for grammatical errors—all of that is fine. The problem is that you can't outsource the act of thinking. Especially for Chinese department students—this is what you study, this is what you'll rely on to make a living in the future. If you let AI do the thinking for you even on your thesis, then what exactly have you been learning for four years?
My old classmate said one last thing to me that I found quite piercing. He said, it used to be that students wrote papers and teachers graded them. Now it's almost becoming: students are the setters, AI is the spiker, and teachers are the referees—seeing who can set in a way that doesn't look like a machine.
That's not a pleasant thing to hear, but it might be true. The first batch of Chinese department students using AI writing assistants may really be losing their ability to write papers on their own. And what's even more frightening than that is that they may not even realize what they've lost.