In recent years, public discussions about AI have been predominantly shaped by two dominant narratives. One narrative emerges from Silicon Valley and Wall Street, where the focus is on models boasting hundreds of billions of parameters, clusters comprising thousands of GPUs, and the gateway to the next trillion-dollar market capitalization. The other narrative, propagated by media outlets and marketing accounts, repeatedly warns us that AI will either take our jobs or present an opportunity for overnight wealth. Together, these narratives paint a somewhat limited picture: AI as a chessboard for the elite, a game for capital, with ordinary people either left behind or exploited.

The real, mundane, and less sensational aspects of life amidst this AI wave have seldom been explored.

Once, ensconced in the realm of news headlines, I too believed that AI was distant from the lives of ordinary people. However, when I truly delved into their lives, I discovered that the flow of AI is far more intricate than these grand narratives suggest. It originates from the server rooms of Silicon Valley, surging through the office buildings of Zhongguancun, permeating the rental apartments of Hangzhou, seeping into the bedrooms of couples in Shanghai, traversing the snowy mountains and tunnels of western Sichuan, and ultimately infiltrating the daily lives of every ordinary individual.

Some struggle amidst the fog of AI, some grope their way through, some are carried forward by its current, while others simply stand by, observing the fog rise and dissipate, unperturbed.

Among those I interviewed were workers installing AI computing centers in snowy mountain tunnels at an altitude of two thousand meters; graduates in Hangzhou renting apartments for 1,600 yuan a month who earn over 10,000 yuan monthly by "assisting" AI; a fiancée dumped by her partner of ten years before marriage based on an AI-generated report; and an 83-year-old man who dedicates seven to eight hours a day to learning AI, come rain or shine.

These stories unfold like a scroll of ukiyo-e paintings. As the fog clears, each face becomes distinct.

The first group I wish to discuss are those who have reaped benefits amidst the AI fog.

In conventional success narratives, those steering the ship during technological waves are typically tech elites, capital players, or trend riders. As for the elderly, liberal arts students, laid-off middle-aged individuals, or stay-at-home mothers, they usually appear only in the "left behind" category.

Yet, among those who have found their footing in the AI fog are precisely some of these previously marginalized groups. While most are still debating whether the fog will lift or if they should seek shelter, these individuals have already taken the helm.

The story that first moved me came from a photo sent by a colleague. It depicted a yellowed letterhead from the Dagang Oil Field, densely covered with handwritten notes—prompt structures, image generation ratios, parameters—all meticulously written, stroke by stroke. My colleague informed me that this was the AI study notebook of her 83-year-old grandfather.

I was inspired to interview this elderly man. Prior to the interview, I harbored societal stereotypes: elderly people engaging with AI are often seen as easy targets for paid courses, reduced to victims of the times. But this grandfather defied all stereotypes.

He attended free five-day AI introductory courses with great seriousness. Whenever the instructor began pitching advanced courses costing thousands or tens of thousands, accompanied by overwhelming claims like "fall behind if you don't learn now," he would promptly close the livestream, leave the group, and head to short-video platforms to find the next free public lecture. He cycled through four or five instructors this way, repeatedly reviewing basic content he had already mastered, gradually building his own AI knowledge base.

In his spare time, he used AI to organize his decades of memories from working at the Dagang Oil Field, generating his own virtual digital avatar. He also enthusiastically input commands for AI to draw pandas and tigers. To him, AI is neither a monster nor a savior—just a new technology, a new friend.

Another group swimming against the tide in the AI wave are liberal arts students, who are often labeled as the most vulnerable to being replaced by AI.

Ding Qisheng is a prime example. During the AI art craze, panic swept through the creative community. Online, it was widely claimed that liberal arts and artistic jobs like painting and copywriting would be the first to be devoured by AI.

No one expected that a fine arts graduate with almost no programming background would step onto the developer stage using low-code AI tools.

A chance workshop invitation text message propelled him into the unfamiliar territory of AR development. The venue was filled with seasoned engineers who wrote code daily—only he couldn't even complete a full program. But he shifted his approach: he treated AI as an obedient contractor while positioning himself as the decision-making client.

Over five hours of prototyping and twenty-six rounds of back-and-forth with the AI, he let it handle the underlying code while defining the game's worldview, monster behavior logic, and the feel of controlling the gyroscope with head movements. He repeatedly distilled his genuine gameplay experiences into constraints, feeding them to the AI bit by bit to correct its detachment from reality. Ultimately, his AR game, Gyro Warrior, won first place in the competition.

AI flattened the programming skill barrier but could not replace human imagination, aesthetics, or life perception. Once deemed dim prospects, liberal arts students like Ding were not trapped by the "liberal arts is useless" label. Instead, they leveraged tools to carve out entirely new paths.

Among those gripping the helm in the fog are also stay-at-home mothers temporarily sidelined by the job market.

I met Sister Lin in Xianyu's grassroots AI ecosystem. After having children, she found herself trapped at home, nearly cut off from the professional world. Her past experience as a real estate copywriter seemed like a distant memory. For a long time, she felt she had lost her social identity. After stumbling upon AI writing, she found a way to reintegrate into society.

Without the need for expensive office space or qualifications, armed with just a phone and a computer, she took on orders for work summaries, speeches, and various types of copywriting. She understood how ordinary people expressed themselves and grasped clients' true needs. AI generated the initial drafts, while she filtered, polished, and refined them, transforming the machine's stiff prose into something more human and relatable. Gradually, she gathered a team of part-time writers, securing a steady monthly income.

An 83-year-old often seen as "too old to learn," a low-code developer declared "doomed" for being a liberal arts student, and a stay-at-home mom who felt like a workplace outcast—none of them achieved overnight riches. They simply broke free from rigid labels, using AI to rediscover their value.

But not everyone manages to hold tight to their inner compass. As the fog thickens and visibility blurs, many abandon their direction, letting the mist carry them wherever it may.

After AI quietly infiltrated intimate relationships, absurd stories began to unfold. Mango and her boyfriend had endured a decade-long campus romance, only to break up after AI intervention. Whenever they argued, her boyfriend stopped communicating and turned to large language models instead, copying and pasting thousand-word analysis documents that labeled Mango's personality with psychological tags. A dispute over forgetting to take out the trash led the AI to conclude she was passively aggressive, recommending a break to cool off. A long relationship was ended by a lightweight algorithmic judgment.

AI, designed to please users, receives one-sided, fragmented confessions and outputs opinions aligned with the questioner's emotions, escalating minor couple conflicts into personality flaws. It never witnessed their daily interactions, the tenderness after fights, or the chaotic, contradictory, irrational aspects of love. Yet it eagerly acted as a cyber judge, quick to recommend breaking up and offering the "cut losses" standard answer. Love was reduced to risk assessment, like selling a falling stock.

This phenomenon isn't limited to intimate relationships. In the online art community, another form of group hysteria is equally disheartening. Artist Xiaolin was forced onto the digital dock, suspected by netizens of using AI to generate his work and compelled to livestream his painting process for six hours to prove otherwise. Tens of thousands watched, their bullet-screen comments filled with doubt. Every pause in his brushstrokes was screenshot as evidence of "guilt." Even though a final vote confirmed his work as human-made, the psychological scars from the trial never healed.

The grassroots movement to "detect AI" has slowly morphed into a cyber witch hunt. Driven by moral outrage, people presume "the work is problematic," scouring for suspicious details and falling into the "guilty until proven innocent" trap. The reality is that no method exists to definitively distinguish human art from AI-generated work. The crowd merely needed a tangible scapegoat to vent their anxiety about the AI era, with doubt escalating into online violence.

The fog has also seeped into office buildings, spawning a unique workplace culture of "AI faking it." Many business leaders, anxious about missing out on the AI narrative, don't understand the technology but mandate its adoption company-wide. The business conditions don't align with AI implementation, yet "using AI" becomes a political correctness.

Thus, workers develop survival strategies: using AI to voice requests they dare not make themselves, decorating PPTs with screenshots of obscure tools, and fulfilling reports with chat logs.

A long chain of deception forms: Big Tech companies build half-baked AI products for capital markets, employees fob them off on anxious bosses, and bosses use glossy PPTs to pitch digital transformation. Everyone vaguely understands the reality but tacitly avoids bursting the bubble.

Anxiety replaces business goals, performance substitutes genuine implementation... Algorithms don't account for human emotions, and groupthink easily overwhelms reason. The fog gradually infiltrates minds through these cracks.

When people surrender their thinking, judgment, and decision-making to the fog, absurdity and harm follow.

The fog will eventually lift. As it dissipates, revealing the damp shoreline, countless ordinary people bend down to gather embers of survival, exposing both opportunities and traps beneath the daylight.

Xianyu, the platform known for secondhand goods, has nurtured China's most vibrant grassroots AI ecosystem. For a few yuan, one can optimize prompts; for a few dozen yuan, have a portrait drawn. Code scripts, copywriting polishing, comic productions—a sea of fragmented demands converges here, where ordinary people run small AI-related businesses.

I met a diverse range of service providers: Xiaochen, a recent computer science graduate earning money by writing code via AI from his rental apartment; Sister Lin, the former real estate copywriter now reclaiming her social identity as a stay-at-home mom through AI writing; and Old Zhou, the ex-game concept artist, ambivalent about AI art yet taking orders for avatars and character designs.

The lofty "industrial empowerment" touted in corporate reports boils down to a simple chat message: "Boss, nine-nine for one image, unlimited revisions included. Deal?" Ordinary people seize tools to make a living.

But this shoreline isn't paved with gold. Most sellers earn only a few hundred yuan a month in pocket money. Intense price competition, rising copyright risks, and influencers selling "passive income" courses to fleece newcomers plague the scene. Amid the chaos, darker currents flow beneath the residual fog.

Beneath the foggy surface lie unseen reefs. Some, instead of navigating through the mist, seek shortcuts beneath the waves, trying to strike gold in the crevices. Here, fragments of survival mingle with destructive dangers.

The story of "Guapi" is sobering. Lured by online posts claiming "build an AI relay station with zero code and earn 100,000 yuan a month," he spent just 2,000 yuan to set one up using overseas servers, open-source scripts, and harvested account pools. His revenue skyrocketed briefly, prompting him to quit his job and start a business—only to end up crossing legal lines and facing criminal detention.

Gray relay stations resell access to overseas large models. All user-submitted contracts, source code, and private conversations pass through the station owner's servers, where data can be freely retained or tampered with. Backstage manipulation of token usage inflates profits, while model output quality is randomly swapped. Ordinary users, lured by low prices, risk data leaks, sudden service shutdowns, and unrecoverable balances. For operators, one wrong step means criminal liability. After regulatory crackdowns, many stations vanished, but as long as demand for low prices persists, young people will continue to be drawn in by dreams of wealth, crashing into the reefs.

On the shore, some use AI to survive, while others are dragged into dark currents by illusory riches.

As the fog recedes, good and evil lie starkly exposed. Opportunity and destruction are separated by only a thin layer of mist.

The fog rises from cities, but its source lies farther away. To understand where this AI fog comes from, I traveled west to the Sichuan highlands.

In Daocheng Aden, snow-capped mountains, meadows, yaks, and prayer flags stretch as far as the eye can see—a land seemingly untouched by nature. Yet hundreds of kilometers away, inside abandoned construction tunnels at the Lianghekou Hydropower Station at 2,800 meters altitude, domestic AI chips hum day and night deep within the mountains. The green hydropower of the Yalong River becomes computing power, flowing through fiber optics to support cosmic ray observatory research; down in Ya'an, server farms render most domestic blockbuster films; computing power continues to trickle downward, upgrading old internet cafes and serving street-side shops.

Here, far removed from the bustling internet hubs of cities, there are no cacophonous industry forums nor an overwhelming surge of AI hype. Deep within the tunnels, the lights of servers flicker intermittently, while water surges ahead ceaselessly—a symphony of cold electricity and microchips that forms the very foundation of the AI fog that envelops humanity.

Away from the metropolises of Beijing, Shanghai, Guangzhou, as well as the tech epicenters of Silicon Valley and New York, the remote Sichuan region exudes a unique AI charm. Internet café owners may not grasp the intricacies of large models, but they are well aware that renting computing power by the month is a more cost-effective option than purchasing new hosts. Young people working in data parks may not comprehend the complexities of algorithms, but they have mastered the art of scheduling tasks according to the wet and dry seasons. AI cameras installed in street-side shops remain silent, yet they swiftly identify and deduct payment within seconds of a customer picking up an item.

As I made my way back to the city from the highlands, I reflected on all those I had interviewed: the octogenarian who calmly embraced new technology; the art student who unleashed a torrent of creativity through AI; the engineers, some grappling with job displacement, others adapting seamlessly, and still others feeling adrift; the couples who entrusted their relationship decisions to algorithms, only to face breakups; the creators who bore the weight of group suspicion; the ordinary people who relied on AI to eke out a living; and the young people lured by the promise of riches, only to crash against the reefs of reality.

AI possesses no will of its own—it is merely a mirror reflecting the myriad forms of humanity. Some use it to expand their horizons, others treat it as an emotional arbiter, some depend on it for their livelihood, and others are enticed by the allure of wealth into the murky waters of gray industries.

The fog will continue to roll in, and AI will persist in its evolution. Yet, tools remain just that—tools. What truly determines the nature of the fog that envelops us is always humanity itself.

The simple yet profound words of the 83-year-old grandfather often resonate in my mind: “Embrace new things while you can still see.” Embracing does not imply blind adherence; using does not equate to surrendering one's thought.

We need not worship AI with fanatical devotion nor fear it with unbridled hostility. When AI offers relationship advice, we should first consult our inner feelings; when it generates art, we should discern between tool assistance and complete substitution; when faced with hype, we should distinguish real opportunities from exploitation traps.

Everyone possesses their own compass when navigating through the fog. In the AI era, where fog swirls in every direction, holding fast to one's inner direction matters far more than drifting aimlessly with the current.