Today, major tech companies around the world are engaged in an arms race over AI products, releasing new iterations of their generative AI tools every few months. A few months ago, China’s Deepseek caused a brief panic in Silicon Valley and Washington by achieving performance close to ChatGPT at a significantly lower cost\u2014this only intensified the competition. Now, ChatGPT, Grok, Gemini, and Llama have once again left Deepseek behind in terms of performance.<\/p>\n\n\n\n
No company wants to be left behind, and most users are eagerly awaiting further technological progress. The European Union has tried to position itself as the global regulator of AI technologies, but this has drawn criticism from tech enthusiasts and U.S. politicians, who argue that heavy-handed regulation will stifle innovation. If the EU is the policy regulator of the global industrial chain, then the global humanities, arts, and social sciences (HASS) community plays the role of intellectual regulator. Yet, at present, most scholars in the humanities are also eagerly keeping pace with AI developments, attempting to leverage AI in their academic research.<\/p>\n\n\n\n
Of course, I count myself among the eager followers. Shortly after ChatGPT launched, I penned a trendy piece on how to use ChatGPT to enhance interpersonal communication. Today, I use various AIs intensely on a daily basis, mostly to deal with practical issues in life rather than academic ones. But I am not without concerns. Nowadays, it seems rare to hear scholars in the humanities express concerns about AI\u2014perhaps because they lack a technical understanding and feel unqualified to make informed critiques, or perhaps their voices are drowned out by the overwhelming tide of techno-optimism.<\/p>\n\n\n\n
Next, I will present my concerns and criticisms about the latest AI technologies\u2014especially generative AI. I argue that the rise of artificial intelligence will be accompanied by a great stagnation of the human spirit, including humanity\u2019s intellectual and artistic achievements. I call this theory the \u201cGreat Stagnation Theory.\u201d I don\u2019t necessarily believe it will come true, but I hope to offer it as a provocation for further thought.<\/p>\n\n\n
From antiquity to the present, from Anaxagoras and Confucius to Beethoven, Einstein, and \u017di\u017eek, major breakthroughs in the human spirit have followed a genius-centered model. I believe this assertion is not particularly controversial.<\/p>\n\n\n\n
Geniuses often attract followers\u2014and sometimes betrayers. Freud\u2019s \u201cSecret Committee\u201d was a group of devoted court guards around him, tasked with defending the truth of psychoanalysis and the reputation of their great teacher. However, Freud\u2019s uncompromising stance toward his own doctrine eventually led to the \u201cbetrayal\u201d of talented minds\u2014Adler, Jung, and Rank. The remaining disciples were mostly mediocre. Even today, psychoanalysis still has many devoted followers. If we include Jacques Lacan, then most psychoanalysis enthusiasts in China today have no direct academic lineage from either Freud or Lacan.<\/p>\n\n\n\n
Generally speaking, the more devout or fundamentalist a follower is, the more mediocre they tend to be.<\/p>\n\n\n\n
Some may object by pointing out that modern research in the natural sciences and engineering seems to be the result of collective wisdom rather than individual genius. For example, the very generative AI we\u2019re discussing now appears to be the product of decades of collaborative work by computer scientists rather than the breakthrough of any one genius. This is why I rolled my eyes when some Chinese media recently called Deepseek\u2019s Liang Wenfeng a \u201cgenius.\u201d<\/p>\n\n\n\n
Thus, natural sciences, engineering, and even parts of the social sciences seem to have transitioned\u2014or are transitioning\u2014from independent to collaborative research. There are no more Einsteins, but we have countless universities, labs, advisors, and graduate students.<\/p>\n\n\n\n
I once participated in a discussion where a young social science scholar argued that those of us in the humanities, still conducting independent research, are \u201cbehind the times.\u201d<\/p>\n\n\n\n
I disagree, though I did not refute him at the time. In the broader humanities\u2014whether philosophy, literature, or the arts\u2014I find it hard to imagine significant achievements being made through collaboration. This was rare in the past, is rare now, and will likely remain rare. Co-authoring a novel is unlikely (though I once attempted such a thing with a friend in middle school), as is co-creating a painting or musical piece on equal terms (such works are almost certainly mediocre), or co-founding a philosophical system. Some collaboration may exist under the direction of one individual, and that can be an exception. Freud\u2019s psychoanalytic movement was such an example\u2014he developed the core theory, while his disciples handled the less critical tasks.<\/p>\n\n\n\n
Even then, conflict is possible. I once took a class taught by a philosophy professor who led a project to write a philosophy textbook, with each person contributing a chapter. He told us that the experience made him \u201cquestion life\u201d and vow never to \u201cjoin a group project\u201d again. He was right.<\/p>\n\n\n
Can AI create a philosophical system like Hegel\u2019s, with dozens of volumes worth reading line by line? Can it write literature on par with Dostoevsky or Milan Kundera? Can it formulate a system like Freud\u2019s psychoanalysis? Can it produce a physics theory like Einstein\u2019s? Can it paint like Raphael? I\u2019m not asking about imitation\u2014I mean the creation of entirely new works or paradigms in similar styles.<\/p>\n\n\n\n
I\u2019m not an AI expert, so I can\u2019t answer in technical depth, but I believe the answer is fairly clear: no.<\/p>\n\n\n\n
More precisely, as of today, it cannot. If it could, we would already see countless genius-level works. While we do see many interesting and insightful things, we haven\u2019t seen anything of that caliber. AI has not yet delivered major breakthroughs of the human spirit.<\/p>\n\n\n\n
From my experience, today\u2019s generative AI is essentially a recombination of existing information. While genius also builds on existing work, it makes astonishing leaps beyond it. AI cannot make such leaps\u2014no matter how sophisticated the prompts. I once asked ChatGPT to create a theory as creative as Freud\u2019s psychological model. It responded with a poor imitation of the \u201cid-ego-superego\u201d framework. I also regularly ask it to generate feature images for my articles\u2014it always churns out formulaic results.<\/p>\n\n\n\n
So here\u2019s a fascinating question: what is it in the human brain that enables true creativity? Whatever it is, current AI doesn\u2019t have it.<\/p>\n\n\n\n
In addition to being recombinative by nature, AI is especially cautious. It emphasizes balance and avoids one-sidedness. Yet Freud\u2019s libido theory, Hegel\u2019s absolute idealism, and Schopenhauer\u2019s will-to-live are all, in some sense, extremely one-sided. And it is precisely this one-sidedness that leads to greatness\u2014whereas excessive balance often leads to mediocrity.<\/p>\n\n\n
In recent years, generative AI has dealt significant blows to the human spirit, and humanity has reacted in a state of panic. Recently, Fudan University drastically cut its humanities faculty and students, while aggressively expanding so-called \u201cinnovative engineering\u201d programs\u2014an overreaction to this shock. Many Chinese universities are promoting AI literacy, often mandatorily or semi-mandatorily. This is strange, considering even top U.S. tech schools haven\u2019t made such extreme moves.<\/p>\n\n\n\n
Some humanities scholars have published a flurry of papers\u2014but most are about \u201chow to better use AI,\u201d not critical reflections on AI itself. There\u2019s a growing sentiment that those who don\u2019t learn AI will be left behind.<\/p>\n\n\n\n
This widespread promotion and adoption is almost certain to result in overreliance. Look at social media and smartphones\u2014many people think they\u2019re using them in healthy, moderate ways, but they\u2019re actually addicted without realizing it. Microsoft thoughtfully named its chatbot \u201cCopilot,\u201d suggesting it\u2019s just a co-pilot\u2014but humans may still end up overly dependent on their co-pilots.<\/p>\n\n\n\n
Previously, people relied on search engines and databases\u2014they were passive tools requiring active human thought to organize information into coherent ideas. Now, relying on AI means outsourcing the thinking process itself, saving mental effort\u2014but at the cost of creativity.<\/p>\n\n\n\n
Many people believe AI can serve as their research assistant while they remain the researcher. It\u2019s like having a reliable grad student to help out. But gradually, that grad student starts coming up with ideas and writing the papers. Initially, AI is supplementary; eventually, it becomes central.<\/p>\n\n\n\n
As we discussed earlier, AI is inherently uncreative. So if AI takes on the main role and humans use it unreflectively, then their work too will be uncreative.<\/p>\n\n\n