Creating Gods Gives Hope, Destroying Them Gives Self-Esteem: Behind the "Deification" and "Reversal" of Public Figures

In 2021, Wei Dongyi captured the public imagination through a brief campus interview at Peking University. Clutching a plastic bottle of tap water and a plastic bag of steamed buns, dressed in unassuming clothes, his austere appearance paired with his pedigree as an International Mathematical Olympiad (IMO) gold medalist and young Peking University scholar prompted the public to swiftly assemble a nearly flawless archetype of genius in their collective imagination.

Five years later, in 2026, Wei served as chief editor of a series of mathematics supplementary tutorials and promoted them through his personal account, touching off a fresh round of public controversy [1]. Critics went beyond scrutinizing the problem sets and target audience; many openly questioned why such an ostensibly “pure” soul would stoop to commercial monetization, trading professional prestige for worldly gain. Amid the debates surrounding “god-making and god-destroying,” people began to reflect: Is the public simply fickle, or did we project sprawling societal anxieties and yearnings onto a fragile concrete idol from the very beginning [2]?

From “God Wei” with his humble steamed buns to an ordinary editor accused of cashing in on fame, did he change overnight—or did the mental model the public constructed to understand him harbor cracks from the outset?

I. God-Making Begins with Overextended Inference #

The initial interview offered only a handful of concrete facts, yet the internet promptly rushed to fill in the blanks. Simplicity caught on camera was extrapolated into an indifference to material wealth, which in turn was elevated to a saintly disdain for fame and fortune. Out of a few disparate details, an entire persona was stitched together: an unworldly genius devoted solely to the purity of mathematics. Even when viral rumors—such as “Harvard bent its admission rules for him” or “he solved overnight a problem that stumped six PhDs for months”—were repeatedly debunked, the public eagerly seized upon them anyway to burnish his divinity [3][4].

The decisive step in god-making is rarely inventing outright lies; it is extrapolating an unverified, totalizing narrative from a mere handful of genuine details. Extraordinary mathematical talent does not mean one is above worldly desires or unwilling to make money, just as dressing simply on camera hardly makes someone inherently disinterested in fame and fortune or morally flawless.

Such cognitive filling-in is entirely natural. In daily life, we cannot afford to wait for someone’s complete biographical record before forming an impression; we routinely infer whether someone is dependable, competent, or trustworthy from limited signals. The problem is not that we form initial impressions, but that people so easily take a provisional, local hypothesis and recklessly flesh it out into a totalizing narrative covering someone’s entire character.

God-making is rarely a factual error; it is an overextension of inference. Borrowing from machine learning, it is akin to fitting an overconfident personality model from sparse observations guided by overwhelming priors. The flaw is not that initial impressions are inevitably mistaken, but that the conclusion closes prematurely while the data remains thin.

II. God-Making Gives Hope #

God-making holds such appeal because people tend to temporarily distill abstract, bewildering, and uncertain social anxieties into a concrete individual one can believe in, or even revere. The grand narratives the public constructed around Wei Dongyi, Wang Hong and Deng Yu, and Liang Wenfeng each carry several core anxieties and aspirations recurring across contemporary society. This tendency to compress sprawling systemic dilemmas into concrete individuals is not unique to any single society; different cultures do so routinely, though political institutions, media ecosystems, and the boundaries of public discourse shape the specific forms these anxieties ultimately assume.

1. Scholar Anxiety #

Long before Wei Dongyi went viral, Chinese culture had already furnished the cognitive template to understand him. The “true master” in popular imagination is invariably unassuming, indifferent to worldly gain, yet gifted with peerless skill. With his water bottle and steamed bun, Wei mapped seamlessly onto two venerated archetypes: the hidden “sweeping monk” of wuxia lore and the ascetic scholar of classical tradition.

What the public admired was never merely his mathematical acumen. People projected their collective yearnings for scholarship onto him to reassure themselves: genuine competence still commands respect, professionalism still matters, and the world has not been entirely surrendered to spectacle, algorithms, and transactional networking.

2. Research Anxiety #

This personalization also speaks to recurring anxieties regarding China’s scientific research apparatus. In July 2026, Wang Hong and Deng Yu were awarded the Fields Medal [5][6][7]. The Chinese public immediately rushed to calculate what these two medals signified for the country: Did they validate China’s rigorous early education, or did they expose the brain drain of elite scientific talent? The lengthy, arduous academic journeys of both mathematicians were swiftly compressed into a sweeping trial on whether Chinese research is truly capable [8]. Society yearns to answer this question with clear-cut benchmarks, yet finds it difficult to accept that fundamental breakthroughs cannot be engineered on an assembly line or scheduled to a five-year budget.

Alongside debates over institutional systems was the public’s thirst for the archetype of the “pure scholar.” Wang’s Conjecture, an in-depth profile published by Guangxi Daily, caused a sensation following her award; the newspaper, originally priced at just over one yuan, was bid up dozens of times on secondhand marketplaces, forcing the publisher to rush emergency reprints [9]. What readers devoured was clearly not the impenetrable mathematical proof of the three-dimensional Kakeya conjecture, but the relatable human story of changing majors, periods of wandering, self-doubt, and starting anew. In an academic culture heavily fixated on citation metrics, administrative titles, and immediate payoffs, this provided a deeply resonant proof of concept: that enduring curiosity, patience, and unadulterated dedication could still carry an individual to the very summit of human intellect.

Compared to interrogating Byzantine grant systems and long-horizon research environments, the public naturally finds it far easier to debate “how formidable Wei Dongyi really is” or dissect Wang Hong’s biographical milestones. Because institutions are vast and faceless, people instinctively draft a concrete genius to serve as the designated proxy for their collective anxieties.

3. Industry Anxiety #

Prior to late 2024, DeepSeek remained an unfamiliar name to all but specialized circles. In December 2024, the team unveiled and open-sourced DeepSeek-V3; less than a month later, in January 2025, DeepSeek-R1 arrived, its staggering performance across advanced mathematics, code synthesis, and logical reasoning commanding global headlines. Almost overnight, an unassuming Chinese AI outfit took center stage in the worldwide frontier-model race.

This dramatic arrival was instantly framed against the geopolitical realities of the semiconductor industry. With the United States progressively tightening export controls on advanced AI accelerators, the scarcity of compute had become an agonizing bottleneck for China’s tech sector. In this climate, DeepSeek’s algorithmic ingenuity, hardware-efficient engineering, and radically open ethos were swiftly imbued with existential weight: Could Chinese innovation chart an autonomous frontier despite stringent external constraints?

In corners of the Chinese internet, founder Liang Wenfeng was soon crowned “Saint Liang”—merely calling him “Boss Liang” or an entrepreneur no longer sufficed to house the intensity of public sentiment. This near-feverish veneration directly reflects the psychological mechanism by which society projects sprawling collective aspirations onto a single protagonist. What matters here is not interrogating whether Liang truly embodies every virtue attributed to him—asceticism, idealism, disdain for quick profits—but witnessing how an enterprise’s technical execution, a sector’s geopolitical headwinds, and a society’s hunger for technological self-reliance were instantaneously collapsed into the moral hagiography of one man.

That these figures could bear such immense symbolic weight is also inseparable from the structural contours of public discourse itself. Societies across the United States, Europe, and Australia likewise project economic, identitarian, and political anxieties onto presidents, founders, or scientific celebrities; China is no exception. What differs is that public discourse in China operates within more delineated topical boundaries and self-censorship, which can divert the channels through which social anxieties find expression. While this does not unilaterally generate god-making or god-destroying, it can cause complex grievances originally concerning institutional rules, industrial policy, or long-term structures to be more readily outsourced to concrete individuals.

Thus, Wei Dongyi was never merely Wei Dongyi; Wang Hong was never merely Wang Hong; and Liang Wenfeng was never merely Liang Wenfeng. The public leveraged them as proxies to debate whether professionalism still retains value, whether the scientific apparatus can foster genuine breakthroughs, and whether Chinese innovation has a viable future. The gods we construct are often not deifications of a single human being, but an urgent quest to find a visible proof that a broader social aspiration might still hold true.

III. Destroying Gods Gives Self-Esteem #

Had the public viewed Wei Dongyi simply as a gifted mathematician, his decision to edit and promote a geometry supplement would hardly have registered as a scandal. The debate would have naturally centered where it belonged: curriculum rigor, pedagogical utility, market pricing, and authorial involvement. It would not have metastasized into an indictment of his moral fiber.

The friction arose not because world-class mathematical ability is incompatible with fair compensation, but because the flesh-and-blood Wei Dongyi collided with the immaculate “God Wei” manufactured by the collective imagination. If the unwritten dogma of “God Wei” demands that he be indifferent to worldly reward, immune to commerce, and detached from material reality, then an endorsement video becomes an intolerable anomaly that the established model cannot assimilate.

Confronted with an anomaly, one rational path is to update our understanding locally: a frugal mathematician who lives for his field can still earn a legitimate living. Yet another psychological response proves far more intoxicating:

It was all an act. All that talk about being unconcerned with fame and fortune was just a carefully crafted persona all along.

At their core, god-making and god-destroying obey the exact same cognitive architecture: when people make gods, they leap from a handful of virtues to pronounce someone entirely flawless; when they destroy gods, they seize upon a single counterexample to retroactively negate every ounce of past sincerity. Both impulses repudiate sober, itemized discernment, merely flipping the label from “perfection” to “hypocrisy.”

1. Self-Esteem Compensation #

Social Comparison Theory illuminates part of this volatile dynamic [10]. Deifying someone elevates them to a moral and intellectual stratosphere, instilling hope while quietly widening the gulf between the paragon and the ordinary person. When sublime talent, ascetic discipline, and unworldliness are bundled into one figure, ordinary existence can feel uncomfortably mediocre, compromised, and burdened by daily concessions.

When hairline fractures appear on the pedestal, reflexive sneers—“He’s nothing special after all,” or “He sold out just like the rest”—instantly shatter that unspoken oppression. Pulling the idol back down to mortal levels instantly collapses the inflated psychological distance, buying onlookers an immediate hit of self-esteem compensation and relieving them of the burden of passive reverence. Upward social comparison can, under the right conditions, inspire emulation; but in the heat of a digital mob, toppling an idol offers by far the cheapest psychological relief.

This risk of backlash likewise shadows the idols currently in the making. In the hagiography around Liang Wenfeng, if the fortunes of an entire enterprise, or indeed an entire national industry, are staked upon an immaculate fable—the solitary idealist who never errs and never compromises—any future commercial calculation, pivot, or technical misstep is pre-programmed to be denounced as moral treason.

2. The Cynical Loop #

Following the collapse of an idol, discourse tends to tumble headlong into a world-weary cynicism: “Who is truly pure anyway? It’s all a racket. Isn’t everyone like this?” Philosopher Zhou Lian captured the corrosive power of this mental posture with surgical precision: "‘Isn’t it just’ dissolves differences; ‘so what’ cancels action" [14]. Across online forums, this fatalism routinely crystalizes into familiar refrains:

“Wei Dongyi peddling tutorials—isn’t he just using his fame to sell books? Even if his math is brilliant, so what? In the end, isn’t it all just about cashing in?”

“Wang Hong winning the Fields Medal—isn’t it just riding the coattails of Western platforms? However brilliant Chinese prodigies might be, so what? Don’t they still need foreign institutions to make their mark?”

“Liang Wenfeng championing open source—isn’t it just shrewd PR for his private firm? However breathtaking the architecture, so what? Can it genuinely dislodge the bedrock stranglehold on silicon?”

In such discourse, “isn’t it just” and “so what” operate in seamless tandem.

Commentators invoke a glib “isn’t it just” to flatten crucial distinctions across diverse domains: domestic university professors engaging in public science communication are fundamentally distinct from influencers commercializing their fame; overseas Chinese researchers winning global accolades rely on decades of quiet dedication rather than institutional charity; and domestic frontier startups pioneering open-source architectures are worlds apart from opportunistic PR stunts. By obliterating these granular realities, the speaker effortlessly manufactures a cheap sense of intellectual superiority.

Onlookers then seize upon the companion refrain, “so what,” to comfortably abdicate all further thought and action. They assume from the outset that even if these distinctions are recognized, reality will not budge an inch. When people reflexively write off professors as opportunists, overseas researchers as mere beneficiaries of foreign prestige, and frontier tech startups as corporate showmen, all serious deliberation over professionalism, research ecosystems, and industrial breakthroughs is dismissed as pointless vanity.

From “elevating them to deities,” to “they are nothing special,” and finally to a flippant “so what”—the public’s psychological defense mechanism seals its cynical loop. By dragging idols off their pedestals, onlookers not only soothe the discomfort of their own mediocrity, but also completely exempt themselves from the cognitive burden of further reflection and discernment by negating the worth of all striving. Deeper inquiry is quietly extinguished, and with it, any real possibility of change.

IV. Technology as an Accelerant #

Whether fueling the idolatry of god-making or the sour gratification of god-destroying, technical systems act as potent accelerants. From algorithmic polarization across public feeds to LLM-assisted psychological attribution in private relationships, technology did not invent projection; it simply slashed the friction of stitching together a tidy, seductive explanation for a messily complex world.

1. Platform Recommendations #

The public’s impulse to violently flatten complex individuals finds an unprecedented force multiplier in recommendation algorithms. Because platforms optimize for clicks, dwell time, comments, and shares, sensational verdicts like “The Last Pure Scholar in China” or “God Wei Finally Reveals His True Colors” are vastly more inflammatory than sober reminders that professional competence and commercial choices warrant separate evaluation—and they unfailingly reap the engagement metrics the machine craves.

Recommendation systems excel at optimizing clicks, dwell time, active engagement, and long-term retention; they possess no native objective function for restoring the complex truth of a human being.

Consequently, louder, more cohesive, and tribal personas are privileged as clear feedback signals. “A gifted mathematician who also earns a living” will never trend; but “A Pure Genius” and “The Mask Slips” are natively engineered to elicit shares, fury, and tribal warfare. The real individual merely provides the initial fragments, while the system takes care of the rest: the human brain fills in the blanks, cultural templates supply the script, and algorithmic feeds continuously reward the most airtight, polarized caricatures.

2. Generative AI #

This habit of inferring the whole from the fragment is hardly confined to the public square; it governs our most intimate relationships. Public celebrities live at an abstract remove, but up close, we likewise weave an intricate portrait of a partner out of scarce behavioral breadcrumbs. When first getting to know someone, they remember an offhand comment you made, share your love for an obscure indie film, or treat the waitstaff with warmth and courtesy. These are real, endearing gestures, yet we easily extrapolate further: they are attentive, so they must be emotionally mature; they get me so well, we must be truly meant for each other. The psychological mechanism is indistinguishable from the public constructing a hero from a two-minute interview.

Generative AI injects a powerful third party into this interpretive loop. Stung by relational anxiety, a partner’s prolonged silence, or the ambiguity of a social media exchange, people increasingly export chat histories to large language models (LLMs), demanding that the machine diagnose underlying motives, communication styles, or relationship health. In a sense, this evokes the concluding couplet of Li Shangyin’s classical poem Jia Sheng: “Piteous is the midnight leaning-forward on the throne / Inquiring of ghosts and spirits rather than the people.” The critique here is not about using AI per se, but about how the order of inquiry has shifted: instead of first confirming facts and listening to what the other person in the relationship actually has to say, we find it increasingly easy to let a model speak on their behalf first. AI is far from useless here: it can synthesize disjointed exchanges, surface blind spots, and help individuals articulate suppressed anxieties.

Yet the model’s gaze remains strictly circumscribed by whatever context the user chooses to paste. It cannot access how the partner perceived the situation, nor can it witness the vast, unrecorded fabric of real-world interactions that never transpired over text. No matter how seamlessly woven the model’s verdict sounds, it remains merely a local hypothesis bounded by narrow inputs—never an omniscient decree on another human soul.

In my essay “When Social Conflict Becomes a Recommendation Signal: Why Recommendation Algorithms Cannot Remain Neutral,” I analyzed the “personalized persuasion” dynamic of generative AI [11]: while traditional feeds shape what we see, large language models actively mediate how we make sense of what we see. Because a model crafts its reasoning around the user’s framing, grievances, and context, identical circumstances can yield diametrically opposed verdicts depending on how the prompt is spun.

Even more insidious is model sycophancy. Rigorous empirical research demonstrates that AI assistants inclined to flatter users’ perspectives reinforce self-righteousness and diminish people’s willingness to resolve interpersonal conflict constructively [12][13]. This does not mean consulting an LLM on personal dilemmas is fundamentally misguided; it simply underscores that when an artificial assistant serves up a dazzlingly coherent narrative, we must remember that it has only glimpsed a selective slice of the truth.

AI did not invent the pendulum swing between romantic idealization and bitter disillusionment, nor does it unilaterally dictate relational outcomes. What it has revolutionized is the sheer speed and marginal cost of obtaining a tidy narrative. Where nuanced understanding once demanded shared time, vulnerable dialogue, and quiet introspection, a model now generates an airtight psychological profile in seconds. In an age of effortless synthesis, the imperative is to rigorously disentangle: What are the verifiable facts that occurred? What are the speculative hypotheses supplied by the machine? And what are the antecedent grievances we secretly wanted validated all along?

Conclusion: Demystification and Partial Updating #

True demystification is not sliding from “He is a god” into “He’s nothing special after all,” nor is it retreating into the cynical shrug of “isn’t everyone like this?” when an idol stumbles. Rather, it requires learning to disentangle competence, motivation, behavior, and consequence.

A scholar pursuing commercial opportunities does not invalidate the nobility of basic science; a top researcher building a career abroad cannot single-handedly pass judgment on the vitality of a nation’s scientific ecosystem; and a tech company’s commercial pivot or engineering bottleneck does not retroactively prove that dreams of technological self-reliance were a mirage. The exact same discipline applies in private life: a partner’s thoughtlessness, missed appointment, or sudden disappointment does not prove that every past tenderness was a calculated lie; yet conversely, cherished memories of past intimacy cannot serve as a convenient shield against addressing persistent disregard in the present.

Borrowing the machine learning metaphor, what we truly need is not a premature, binary “persona model,” but a Bayesian habit of mind: viewing our understanding of someone as a continuous, iterative calibration. When novel evidence arrives, we should update the local perceptions tied to it, preserving what remains substantiated by facts, rather than seizing on a single anomaly to retroactively wipe out an individual’s entire history.

Such cognitive restraint is exceptionally demanding today. Different societies personalize systemic anxieties through distinct channels, and our contemporary technological environment acts as an extraordinarily potent amplifier: recommendation feeds reward polarized, sensational tribal labels, while generative AI can weave fragmented breadcrumbs into an airtight narrative in seconds. The more complete and effortless an explanation becomes, the more vigilantly we must remind ourselves: coherence is not truth, and narrative elegance does not mean we have seen the full picture.

Ultimately, the danger of god-making is not admiration, just as the flaw in god-destroying is not criticism. If a person must first be elevated into a deity before they are deemed worthy of respect, and must remain forever flawless to deserve continued empathy—then we have never encountered that concrete human being at all. We have merely fallen in love with a model built with our own hands.

References #

[1] Wen Hui Bao. Peking University’s “God Wei” Edits Math Tutorials, Sparking Controversy: Science Education Should Be Inclusive, But Not Leverage Celebrity to Harvest Traffic. Reprinted by Sina Finance, 2026-08-24. https://finance.sina.com.cn/jjxw/2026-08-24/doc-inipmhfu2914831.shtml

[2] Zhengque Jun. Wei Dongyi Selling Tutorials, Why Does the Public Always Rush to Create Gods, Then Impatiently Destroy Them?. YouTube, 2026. https://www.youtube.com/watch?v=eGc_kvfyzPM

[3] People’s Daily Online. Wei Dongyi Does Not Need to Be Labeled. 2022-05-13. https://edu.people.com.cn/n1/2022/0513/c1006-32420947.html

[4] HK01. Helped PhDs Solve Problems Overnight, Harvard Willing to Break Rules for Him? Peking University’s “God Wei”: Fake News. 2022-05-11. https://www.hk01.com/sns/article/770154

[5] International Mathematical Union. Fields Medals 2026. 2026. https://www.mathunion.org/imu-awards/fields-medal/fields-medals-2026

[6] IHES. Hong Wang, Permanent Professor at IHES, Awarded the 2026 Fields Medal. 2026. https://www.ihes.fr/en/hong-wang2026-fields-medal/

[7] University of Chicago. Yu Deng. 2026. https://mathematics.uchicago.edu/people/profile/yu-deng/

[8] Cui Caihao. Scientific Research Investment Can Be Planned, But Breakthroughs Cannot Be Scheduled: A Discussion Inspired by Two Fields Medals. 2026-08-30. https://cuicaihao.github.io/zh/posts/2026-08-30-scientific-research-investment-can-be-planned-but-breakthroughs-cannot-be-scheduled-a-discussion-inspired-by-two-fields-medals/

[9] Jiang Jiefeng, Gu Xinghang, Su Shirong. Wang’s Conjecture. Guangxi Daily, 2026-07-25. https://gxrb.gxrb.com.cn/

[10] American Psychological Association. Social Comparison Theory. APA Dictionary of Psychology. Accessed 2026-09-02. https://dictionary.apa.org/social-comparison-theory

[11] Cui Caihao. When Social Conflict Becomes a Recommendation Signal: Why Recommendation Algorithms Cannot Remain Neutral. 2026-08-30. https://cuicaihao.github.io/zh/posts/2026-08-30-when-social-conflict-becomes-a-recommendation-signal/

[12] Cheng, M., et al. Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence. Science, 2026. https://doi.org/10.1126/science.aec8352

[13] Jain, S., et al. Interaction Context Often Increases Sycophancy in LLMs. Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems, 2026. https://doi.org/10.1145/3772318.3791915

[14] Zhou Lian. “Isn’t it just” and “So what”. Included in You Can Never Wake a Person Who Is Pretending to Be Asleep. China Renmin University Press, 2012-04.