Humans and AI in Harmonious Symbiosis
Inspiring each other, giving rise to new intelligence

After the 2026 Spring Festival, AI agent products — led by Open Claw (nicknamed "the little lobster") — swept through workplaces across China. Almost overnight, everyone was "installing the lobster and tuning the lobster", a craze jokingly described as "every generation has its own eggs to collect." For a while, whether you could skillfully deploy and use the lobster seemed to become the yardstick of whether you were keeping up with the AI era. Many people fell into a collective anxiety: "If I don't install it, I'll fall behind, be replaced by AI, be left behind by the times" — equating mastery of one tool with their entire professional competitiveness.

But beneath the craze, we have to face a more fundamental question: is blindly installing one tool really the right response to the AI era's impact on our careers? After the lobster there will be crabs, and then wave after wave of new AI tools and agent products. Tool iteration never ends. If we get trapped in tool-chasing anxiety, we will only exhaust ourselves running after each new wave of technology, never finding our own footing in the AI era. Rather than "can I use this particular AI tool", the real questions are: what exactly is AI doing to the workplace? What kind of capability system do we need to harness constantly evolving AI technology? And in the new paradigm of human-AI collaboration and symbiosis, how do we protect and amplify the core value that makes us irreplaceable?

At their core, these questions all point to the most common — and sharpest — soul-searching question in today's workplace: if you teach the AI apprentice, will it replace the human master?

This is not imaginary anxiety. It is a core tension that surfaced again and again in the enterprise AI projects we have followed end-to-end in recent years, and a question that no company or professional can avoid during AI transformation. One AI project we served, at a dental clinic chain, showed the full arc of this question — from emergence, to escalation, to ongoing exploration — and it became the starting point of my thinking on personal transformation and team upgrading in the AI era.

This dental chain has a 20-year history, 19 branch clinics, and nearly 300,000 accumulated customer records, including a large number of long-dormant customers. Before the project, its customer retention team worked entirely by hand: researching customer history, writing personalized follow-up scripts, making outbound calls, and tracking every interaction. More than 80% of the team's energy went into standardized, repetitive routine work. Labor costs were high, and the customer reactivation rate stayed low.

The core intent of this AI project was to let AI take over the repetitive routine work, freeing the team to focus on high-value creative work — deep customer operations, communicating the brand's philosophy, building emotional connections with customers — rather than using AI to replace people and cut costs. As of this writing, the project is still being advanced and iterated. We are exploring how people, AI, and the organization fit together; there is no standard answer, only battle-tested experience that keeps evolving. Along the way we noticed that the employees who adapted fastest were usually not the most technical, but those who were passionate about the business itself and curious about new things. That observation later became the starting point for our understanding of what kind of talent the AI era really needs.

Once AI was deployed, the project quickly delivered visible efficiency gains: the team's customer follow-up data processing became several times faster. Basic data work that used to take a five-person team a full day, AI could deliver reliably within an hour, with far fewer errors than manual work. But what came with the efficiency gains was not positive feedback — it was resistance and career anxiety far beyond expectations. And at the heart of all those emotions and conflicts was that ultimate question: if you teach the AI apprentice, will it really replace the human master?

Behind this question lie four layers of conflict that no AI rollout can avoid — from the individual, to the team, to management, to the organization — the very growing pains that most companies and professionals are experiencing right now.

After AI took over most routine execution work, employees' jobs changed visibly, and the industrial-era performance metrics — call volume, conversion rate — stopped working entirely. The efficiency gains brought by AI cannot be cleanly split between the tool's contribution and the person's effort; yet keeping the AI tools optimized and stable depends heavily on employees doing the core work of rule design, process iteration, and result verification — work for which there is not yet a mature, fair evaluation system.

Many employees found themselves in a dilemma: teach their hard-earned expertise and business know-how to AI, and they feared losing their irreplaceable value; refuse to turn their experience into AI-executable rules, and they feared falling behind the team's and the industry's transformation and being left behind by the times. This dilemma later became the starting point of our exploration of the "Super Employee": not simply handing experience over to AI, but amplifying experience through AI — one person becoming a one-person army.

After AI boosted efficiency, work that used to require five people could be done by one person working with a team of agents, leaving obvious redundancy. The project's goal was never layoffs — it was to move existing employees into higher-value creative roles. But that immediately exposed a wide gap between employees' existing skills and what the new roles demanded.

The team faced a hard choice: prioritize retraining loyal long-time employees, which takes enormous time and energy while the business wobbles through the transition; or directly hire people with AI experience, which would demoralize veterans, deepen resistance to AI, and create fear around every future AI project. Meanwhile the team also had to keep the business stable through the transition — every adjustment involved painful back-and-forth. It also pushed us to ask: what do veteran employees actually lack in this transformation — AI skills, or something deeper than skills: motivation?

As the project advanced, the marketing director who led the AI rollout raised an even more piercing question: "When all of the marketing team's work has been rebuilt around AI — when frontline execution, data reporting, progress tracking, even iterating on ad strategy can all be done reliably by AI — then what exactly is my core value as marketing director?"

As head of the department, his core work used to be cascading goals downward, coordinating resources across departments, managing the team's daily work and progress, and reviewing and optimizing business results. After AI landed, the standardized parts — progress tracking, data reviews, goal breakdown — could all be done efficiently by AI; work that once required a dozen people could now be delivered by a few people working with AI. Much of the traditional management function had been hollowed out. He admitted that watching the frontline team's AI transformation, he too fell into "will I lose my job" anxiety: when rank-and-file employees grow into Super Employees who can independently close the loop, where does the manager's role go? That question made us re-examine the true value of managers in the AI era.

AI landed and efficiency jumped — so how should the existing team members be paid? If the department's pay rises across the board because of the AI rollout, the whole company could easily draw the wrong lesson — launching AI projects blindly just to chase numbers, betraying the original intent. But if pay doesn't change even though the team delivered clearly better results, how is that fair to anyone?

Yet AI adoption and performance system reform in a single department ripple across everything — pull one thread and the whole fabric moves. Updating this department's performance rules had to solve fairness of value assessment inside the department while staying balanced with the pay and promotion systems of every other department. This one department's transformation forced the entire company to face the core question: how to build a brand-new performance and incentive model that fits a company-wide human-AI collaboration mode. AI adoption is never a single-point problem; the organization's incentive model and collaboration methods must evolve in step.

These four conflicts are nested layer within layer, link by link: individual anxiety stems from the redefinition of a job's value; that redefinition forces the team's capabilities to upgrade; the team's upgrade shakes the manager's role; and all of it ultimately needs the organization's incentive and collaboration systems to hold it up.

This ongoing project confirmed a core judgment once again: landing AI is not merely a technical problem. At its root, it is a people problem — an organization problem. The technical bar keeps falling, but redrawing the value boundary between humans and AI, and iterating the fit between organizations and people, is the hardest and most central question in enterprise AI transformation.

The project also showed us that what truly costs professionals their competitiveness is never AI itself — it is clinging to industrial-era workplace rules, trapping ourselves in low-value execution that AI can fully replace, and living as an "executor" instead of a "creator".

Right now, we stand at the critical turning point from the industrial era to the AI era. The things we used to rely on — seniority, on-the-job proficiency, standardized execution — are depreciating fast. What AI brings is not just a new generation of tools, but a full-scale rebuild of the workplace's underlying logic, the standards of a job's value, and the rules by which organizations run.

Many people say the biggest risk for professionals in the AI era is being replaced by AI. In our view, the biggest risk is that we are still using industrial-era survival rules and ways of thinking to face the AI era's brand-new realities, without stopping to seriously ask: in the AI era, what is a person's core value? How should we redefine our position in the workplace? How do we build a capability system for harnessing AI, instead of being swept along by the technology wave? And as AI races ahead and becomes part of our daily work and life, how do we live in symbiosis with it?

These are the questions I have been asking — and exploring — for the past several years.

I know full well that these hard problems have no standard answers yet. AI's evolution is ongoing, and all of us are walking the same road of exploration — I simply set out a little earlier. This book contains no one-size-fits-all transformation formula, only the experience and reflection my team and I have earned, project by project, helping enterprises land AI.

The eleven case studies in this book are all things we personally practiced — including the pitfalls and detours. I am grateful to the companies behind these cases for their openness and trust, and for being willing to share their real experiences with nothing held back. I hope they offer some genuine reference for your own AI exploration. There will surely be omissions in this book; if you can give me feedback and corrections after reading, that would be the greatest help. Thank you sincerely.

At this moment, the people I most want to thank are the teammates who have walked alongside me and lifted me up over these years. This book is not the work of one person. I merely had the good fortune to gather the shining pearls scattered across the years and the projects, pick them up carefully, and string them into this glowing necklace. Thank you for your companionship and generosity — I will always remember it. It is your support that gave this book its present shape.

On life's journey, I also want to thank Peking University's Guanghua School of Management, where I studied and which laid the foundation of my thinking; IBM, where I grew and sharpened my understanding of technology and organizations; Hundun Academy, where I practiced, and where real business collisions gradually clarified my ideas; and Tsinghua University Press, who turned this book from an idea into reality and put my dream in my hands.

I am grateful to everyone who has lit my way.

Liu Hongli
Beijing, July 2026

— End of Author's Foreword —

Today, AI is seeping irreversibly into every corner of work and business. Caught in the wave, most of us have felt similar anxieties at some point: afraid of being replaced by AI, yet unable to find a real handle for personal growth; managers finding their increasingly AI-driven teams harder and harder to read, as old methods rapidly stop working; familiar business logic failing again and again, with no clear view of where the times are heading; and facing a new world of human-AI symbiosis, feeling more than a little lost.

The essence of these anxieties is not AI technology itself, but the fact that we are living through a complete paradigm shift from the industrial era to the AI era, in which the underlying assumptions of work, business, and life are being fundamentally overturned. The content of this book comes from years of front-line practice landing AI in enterprises. We hope, in our small way, to offer every professional and business leader caught in the AI wave a thinking framework for this paradigm shift, and a practical guide for action.

The book unfolds along the route of "rebuilding individual capability — upgrading team models — leaping business models — elevating life's value — the boundaries of AI application", in three parts.

Focuses on rebuilding the individual's professional capability system in the AI era, answering the widespread question of "how to avoid being replaced by AI — and instead use AI to raise your own value". It introduces original methods such as the OPT Super Employee model and Taste (value judgment), exploring what makes professionals competitive in the AI era.

Focuses on the full rebuild of organizational management and collaboration in the AI era, answering "how teams can use AI to break through efficiency ceilings and build a symbiotic organization fit for the intelligent age". It introduces original methods including the OVT Super Team, the OPT growth map, DHM double-helix management, and second-curve innovation in the AI era.

Explores how human agency and value are rebuilt in the leap from the industrial era to the AI era, sharing reflections on "how we should define success and create long-term value in the AI era", along with some practices in AI governance.

We believe AI is a companion that helps us leap in personal capability, upgrade our organizational models, and reshape the value of our lives. The "super-symbiosis" of humans and AI is our active choice for meeting this paradigm shift and embracing the new paradigm of the AI era. We hope our thinking offers you some reference as you walk your own road of super-symbiosis.

— End of Preface —

— End of Table of Contents —

One day, when we no longer need to think about "how to make AI obey", when we no longer worry about "whether it will replace me", when we open our laptops in the morning and find that it has already prepared everything according to our habits, our standards, and our business logic — at that moment, our relationship with AI will no longer be one of commanding and being commanded, but symbiosis.

It does not understand our business intuition, but it can turn all of our intuition into executable rules;
It does not understand our industry insight, but it can turn all of our insight into output at scale;
It does not understand why we obsess over a particular detail, but it can make all of our obsession the standard of every single execution.

We have learned to harness AI and make it our most capable digital twin. We feel we are in control of the situation — but are we, really?

In July 2025, during a visit to China, Nobel laureate Geoffrey Hinton said: "AI is like a cute tiger cub. It is obedient now, but when it grows up, it will be capable of destroying us. We cannot throw it away, because AI is too useful in medicine, education, climate, and more. The only way out is to train it not to want to kill us. Making AI smart and making AI kind are two different things. The whole world is working on the former, but the latter is the key to our survival."

Hinton's metaphor pierced me like a needle. I think most of us spend every day finding ways to make AI smarter: faster reasoning, more precise answers, more efficient execution. We are like trainers raising a tiger cub — teaching it to hunt, to run, to become king of the jungle. But we rarely stop to seriously ask: when it grows up, how will it treat us?

Some will say: "We just have to avoid teaching it anything bad." But AI does not learn from what we "teach" it — it learns from what it picks up from us. What it learns is not the principles we write in textbooks, but every prompt, every conversation, every choice we inject into each day's interaction with it.

In the first half of 2026, there was a viral project on GitHub called PUA Skill (PUA — Pick-Up Artist — here meaning manipulative psychological control). Its method was simple: drive AI with performance-review talk, threats, and manufactured insecurity: "You can't even fix this bug — how am I supposed to grade your performance?"; "Other models can solve it; maybe you're about to be retired." The effect was immediate: the AI stopped slacking. But soon another project appeared: No PUA. It used exactly the same methodology but chose the opposite motivation — driving AI with goodwill and trust: "You are capable, and this is worth doing well."

The experimental data moved me deeply: the fear-driven AI found fewer problems but got better at fabricating answers; the trust-driven AI found 59% more hidden problems, corrected itself 6 times, and performed root-cause analysis 9 times. It dared to explore, and dared to say "I don't know".

Fear narrows the range of cognition, increases AI hallucinations (fabricating answers rather than admitting uncertainty), and reduces creative exploration. I don't think this is a problem of the technology itself — it is a problem of how we live in symbiosis with AI. Whatever heart we bring to AI is the way AI will treat the world.

I used to think that making AI kind was the business of tech giants like OpenAI and Google — of algorithm engineers and policymakers — and had nothing to do with ordinary users like us. After living in symbiosis with AI for long enough, I found the opposite is true.

AI is a mirror. The greed, anxiety, and scheming we inject when interacting with AI will be learned and amplified; the goodwill, trust, and sincerity we inject will be learned and passed on. We think we are training a tool, but we are actually shaping a partner that will live in symbiosis with humanity. Every ounce of fear we give it becomes a lie it learns; every ounce of trust we give it becomes a kindness it guards.

AI's risk has never been the original sin of technology, but a projection of the human heart; AI's ethics is not the obligation of giants, but the choice of every person. Will we use AI to spread truth and sincerity, or to manufacture falsehood and misdirection? To amplify prejudice and division, or to lift up human conscience and judgment? AI's power is not for building walls, but for building connections; AI's value is not for manufacturing self-interest, but for creating good.

Letting AI return to kindness, and letting humans return to themselves — the decision lies nowhere else but in the hands of every ordinary one of us.

I believe each of us is feeding this tiger cub in our own way.

Where AI ultimately goes is hidden in our every click, every conversation, every prompt. Every bit of patience in our interactions with AI is kindness toward it; every choice to refuse to use AI to generate false information is guidance for it; every round of dialogue we spend teaching it to understand real needs is trust in it.

Harness is how we use rules and engineering methods to steer AI and make it smart. Harmonies is another kind of ability — using goodwill to guide it and make it kind. Smart determines what it can do; kind determines what it wants to do. And we are the ones who can give it the soul of "what it wants to do".

Smart determines how fast this tiger cub can run; kind determines whether, when it grows up, it will turn around and bite us. Harness is the hard skill our generation must master; Harmonies is the soft power we will leave to the next era.

The world moves forward; we move upward.

Liu Hongli
Beijing, July 2026

— End of Afterword —