AI's Dual Kill Lines and the Search for Ways to Avoid Them
title: "AI's Dual Kill Lines and the Search for Ways to Avoid Them" date: "2026-09-17" author: "Zhigeng" channel: "frontier" excerpt: "Those who cannot use AI will be killed off by those who can; those who misuse AI will be killed off by AI itself. The first three industrial revolutions replaced physical strength yet always found compensation; the fourth revolution replaces intellectual strength itself, yet the newly opened window is nowhere to be seen. This article dissects this dual mechanism and searches for two moats." tags: [] readTime: 16
This article aims to analyze the AI anxiety that is currently widespread, and to lay out my own exploration of related paths of escape, in exchange with readers and experts.
The "dual kill lines" of AI in the title mainly refer to this: those who cannot use AI will be "killed off" by those who can use AI; those who misuse AI will be "killed off" directly by AI.
I. On the Term "Kill Line"
"Kill line" is originally a term from esports, referring to the threshold of skill at which a player can instantly defeat an opponent once their operation reaches a certain critical point. It is a blood line—a defeated player can revive if they can restore full health, but if their blood runs out, they die.
The game's setting is only the source of the meme. The term "kill line" went viral a while ago because someone used it to refer to certain middle-income groups in American society who, once unemployed, could be dragged into a chain of debt and be unable to recover.
This article borrows the concept to depict the situation in the AI era in which people lose their ability to work or their cognitive ability, and are thereby marginalized by society or abandoned by the times.
This line divides people into two broad groups and four types.
The first group: immigrants of the pre-AI era
If we take the release of GPT-3.5 in November 2022 as the marker of entering the AI era, then those who had entered university by that point can roughly be counted as the dividing line between "pre-AI era immigrants" and "AI era natives." In terms of birth year, this corresponds roughly to 2005. That is to say, for the sake of narrative simplicity, let us tentatively regard the "post-05 generation" as AI era natives, and all age groups born before 2005 as pre-AI era immigrants.
In this way, the immigrant army includes everyone born from the last century to the early years of this century. Some have wittily divided these people into "green," "middle," "old," "divine," "ancient," and "primordial" tiers, to vividly describe the attitudes of different age groups toward new things.
Although the sub-groups differ greatly, these immigrants share one commonality: their education before graduating from university (or, for a small transitional group, high school) did not involve AI. If the education they received was relatively ideal, we can roughly assume their foundation is solid.
Those above the line: carrying the knowledge structure and cognitive abilities formed in the "mother era," they can quickly learn, even skillfully harness, AI tools, treating them as "inorganic organs" that extend their own intelligence. They bring into play the compound interest of their original cognitive framework and thinking foundation, realize human-machine co-evolution in the new era, and can hope to walk to the forefront of the times.
Those below the line: because they have no willingness or no method to master AI technology, or because they cling to the old and reject the new, they miss the opportunity to connect with the new era, or fail to acquire the ability to connect with it. They may be able to hold on for a while within the old system, but will most likely be gradually marginalized in professional competition and social distribution, and may even be pushed out of the advancing ranks of the times.
The second group: natives of the AI era
Those above the line: with high-quality family education or school education, they have promptly made up the lesson of basic knowledge and cognitive framework. They can understand, relatively well, matter, the universe, human society, consciousness, and thinking, and can see clearly the relationship between the real atomic world and the virtual bit world. They can write and express themselves skillfully, are adept at using AI, and become the trendsetters of the times.
Those below the line: having failed to establish the necessary knowledge structure and cognitive framework, they over-rely on AI, leading to their own cognitive degeneration, and are ultimately backfired upon by AI, losing the ability to think independently and becoming "social giant babies," most likely unable to control their own destiny in the future.
II. The Mathematical Community Warned First
Mathematics, as the pearl on the crown of human intelligence, was, in the era of large-scale industrialization, kept at arm's length because it was far from frontline occupations and concrete life. At the turn of the century on the eve of AI, mathematics began to enter the public's serious attention because of its remarkable achievements in actuarial science and modeling. In an instant, mathematics majors became popular and were widely favored in the hot fields of finance and IT, so much so that many people regretted not having chosen mathematics as their major.
In 2024, Fields Medal winner Terence Tao published a paper in Nature, pointing out: "AI is changing the paradigm of mathematical research. Mathematicians who cannot effectively collaborate with AI will soon be eliminated from core research areas."
Tao's judgment is not alarmist. AI can already automatically generate mathematical conjectures and provide preliminary proofs; traditional manual computation and theorem derivation are being replaced one by one by AI tools. The core ability of human mathematicians in mathematical research is shifting from "derivation ability" to "question-asking ability" and "AI collaboration ability."
The reversal came quickly—just a few days ago: Tao again cried out that AI could cause humanity to lose a generation of mathematicians—because AI has deprived young people of the opportunity to grow up in the world of mathematics—and together with 25 other mathematicians, he called for a global slowdown in the AI field, warning against AI's erosion of human cognitive abilities.
Their worry is not without reason: when students can use AI to directly obtain answers to math problems, and when researchers rely on AI to generate paper frameworks, humanity is losing its most basic opportunities for thinking training.
The consequence of this "cognitive outsourcing" is the degeneration of humanity's own cognitive abilities.
More sobering still is that the mathematical community's warning is only a prelude. AI is penetrating all fields that require intellectual labor at an even faster pace—from legal document writing to drug research and development, from news reporting to artistic creation—AI is replacing humans in completing more and more work that originally required deep thinking. In particular, it is sweeping through entry-level positions across all fields, causing a cliff-like reduction in job opportunities for new graduates.
How should the mathematical community collaborate with AI, so as to bring AI's powerful role into play while preserving human creativity and the transmission of the fine mathematical tradition? Mathematicians have not yet found a consistent approach.
The mathematical community is a microcosm of society as a whole. How to find an appropriate way to apply AI has become a thorny task facing all fields of human society.
III. The First Few Industrial Revolutions: The Compensation Mechanism of Physical Tools
Looking back at history, humanity has experienced three major industrial revolutions. In each revolution, tools replaced some human ability, but each also opened up a new space of ability, forming a hidden yet stable chain of compensation.
In the first industrial revolution, steam engines and textile machines replaced the physical strength of manual laborers. Those who lost their manual jobs turned instead to learning to operate machines and maintain equipment, and found their places again in more complex positions. Physical strength was replaced, but the ability to learn mechanical knowledge became the new pass.
In the second industrial revolution, electricity and assembly lines compressed repetitive physical labor further into the rhythm of machines. Correspondingly, society needed a large number of technical workers, engineers, and managers, and the balance of the division of labor tilted clearly toward intellectual work for the first time—you no longer had to earn your living by physical strength alone, but you had to be able to read blueprints and coordinate processes.
In the third industrial revolution, computers took over the heavy intellectual chores of computation, data processing, and network linking. Many of humanity's best talents threw themselves into the design, production, application, and development of new functions of computers and network facilities; hardware and software talent, electronics and data engineers became the pride of the times.
Looking across the first three revolutions, a common pattern emerges: physical tools did weaken human physical strength, but people could always preserve basic physical fitness through exercise and achieve career transformation by learning new skills, ultimately winning their place again in social competition through intellectual work and higher-level labor.
God closes one door, but opens a window. And this is precisely the greatest suspense of the fourth revolution: this time, what is replaced is the "door" of intellectual strength itself, while the newly opened "window" has yet to be seen.
IV. The AI Era: The Uncompensated Dilemma of the Intelligence Revolution
The intelligence revolution of the AI era is fundamentally different from the previous three industrial revolutions: this time, AI replaces human intellectual labor, and we have not yet found an effective compensation mechanism.
The predicament of "pre-AI era immigrants"
As discussed earlier, the so-called "pre-AI era immigrants," apart from a transitional group, roughly refers to the generation that completed their studies (generally measured by university graduation) and entered society before AI arrived. The challenge they face is how to adapt to the universal application of intelligent tools within their existing knowledge structure and cognitive habits.
A person who cannot use AI tools will soon be left far behind by those who can in terms of work efficiency. Many traditional white-collar positions are being replaced one by one by AI automation. On the balance of social distribution, those who cannot collaborate with AI will gradually lose their competitive edge and be squeezed out of the ranks of the times bit by bit.
Where is the "moat" of this group?
The answer: actively embrace the new era, immediately start learning AI capabilities and persist in lifelong learning, and acquire the "feel" of collaborating with AI.
The "feel" here refers not only to proficiency in operating AI tools, but also to the tacit quality of understanding the boundaries of AI's abilities, judging the quality of AI's output, and guiding the direction of AI's creation. A person accustomed to using AI can tell at a glance which AI tool is more handy, whether the AI is operating normally and where a problem may lie, where the AI's output is nonsense, where it is trustworthy, where it is lacking, and how it should be improved. This "feel" cannot be taught by any tutorial; it can only be honed through one real encounter after another.
The risks of "AI era natives"
The so-called "AI era natives" refer to those born into an intelligent environment, or those born in the old era but who grew up and studied in the new era. To them, AI is never a new tool, but something as taken for granted as air and water. Precisely because of this, the risks they face are more hidden, and more fatal.
This risk is called "cognitive outsourcing."
Over-reliance on AI to obtain information and solve problems will cause a person's memory, logical reasoning ability, and critical thinking ability to quietly degenerate. Habituated to the "optimal solution" given by AI, a person will gradually lose the ability to make independent judgments and complex decisions, and then lose the ability to discern material and raise questions. When AI can replace most of humanity's intellectual labor, humanity's unique value will gradually drain away—this is a double crisis: losing both the power of thinking and acting, and the meaning of existence.
Where is the "moat" of this group?
Answer: in establishing a solid basic cognitive framework and acquiring the ability to independently understand problems, make connections, make judgments, and make corrections. And the establishment of this framework and the acquisition of this ability must be completed at the basic-education stage—it absolutely cannot be delayed until after entering university or the workplace before cramming it in.
Because after entering university or the workplace, one needs to fully leverage AI's advantages and collaboratively complete more important learning and work tasks; there will no longer be the time or opportunity to make up for it.
Thus, the question naturally points to education—how exactly should we teach, so that the next generation neither misses out on AI nor is devoured by it?
V. Redefining Education
In the pre-AI era, a person's cognitive growth usually required completing a long, complete cycle of "primary, secondary, and tertiary basic education, plus higher education." But in the face of AI, this "cognitive acceleration tool," we must redistribute the educational functions each stage should bear.
The basic-education stage
Before graduating from middle school, a child needs to complete several crucial things.
First, establish a cognitive framework. Understand matter, the universe, society, human minds, the foundation of values, and the operating mode of AI itself. These seemingly grand propositions can actually be planted bit by bit in the daily life of basic education—physics class teaches the laws of matter, history class teaches the changes of society, biology class teaches the logic of life……
Second, master basic abilities. Hone written and oral expression, learn basic thinking and communication, be able to keenly capture problems, clearly define them, and then accurately express and interpret them; and gain preliminary experience of collaborating with people and with AI. These are the most valuable abilities for later collaboration with AI.
Third, cultivate critical thinking. Learn to question, analyze, and judge, without blindly accepting the answers given by AI. Only a child who habitually asks "why" and can judge right from wrong will not be led by the nose by AI.
Fourth, understand the human-machine relationship. Understand that AI is a facilitator rather than a replacement, understand AI's working mechanism, master the basic principles and methods of collaborating with AI, and acquire that precious AI "feel" and "sense of collaboration."
The higher-education and later stage
In an era of rapid development and rapid decay, universities no longer have the environment for slow, gentle growth. The focus of education should shift from "knowledge transmission" to "ability enhancement," fully leveraging AI's supporting role, and entering a new stage of comprehensive shared progress and growth with AI.
On the foundation laid at the basic-education stage, use AI tools to deeply and efficiently learn professional knowledge and skills; treat AI as a partner in learning and innovation to jointly solve complex problems; use AI to integrate knowledge and resources across disciplines, building one's own unique knowledge network and capability architecture; and at the same time, cultivate the habit of continuous learning to adapt to the rapid iteration of AI technology, keep oneself from falling behind, and ultimately take on the responsibility of the sustainable development and perpetuation of human civilization.
At this point, a clear path of education has emerged:
Basic education is responsible for "becoming human"; higher education is responsible for "growing together with AI."
VI. Recommendations for Parents and Educators
Facing the challenges of the AI era, parents and educators need to complete a fundamental transformation of philosophy. The core of this transformation can be summed up in one sentence: first become "human," then learn to use AI.
We must help the natives of the AI era, before they come into contact with the bit world, to first come into contact with the atomic world and the carbon-based society.
Start from the physical world. Let children first touch, observe, and truly experience, read books written by humans before the AI era, and establish a basic knowledge structure and cognitive framework.
Establish real knowledge connections and social connections. In interacting with real people, learn emotional communication, cooperation, and empathy.
Go understand the human mind itself. Through philosophy, psychology, and other disciplines, understand how humans think and make decisions—why we have biases, why we act impulsively, and why we can restrain ourselves at critical moments.
Appropriately engage with the virtual world. Only after establishing a solid cognitive foundation should we introduce AI tools and virtual environments. At that point, AI is no longer a black hole that devours everything, but a suitable partner for collaboration.
This is equivalent to adding a period of "childhood" for human offspring, using this extended time to understand oneself and understand the world.
Specifically, we must hold fast to four things.
First, curiosity and the desire to explore. This is the source of human creativity, and something AI still does not possess. A child full of curiosity about the world, even in the face of the most powerful AI, will not be easily replaced.
Second, willpower and resilience against setbacks. In such a rapidly changing era, these qualities are very important.
Third, values and ethical outlook. Let children possess a firm foundation of values, clearly understand the boundary between humans and AI, understand what should be decided by humans, and avoid being assimilated by AI's algorithmic logic.
Fourth, aesthetic judgment and decision-making power. This is a unique human spiritual experience, and also an ability that AI still finds difficult to truly understand, and even more difficult to truly possess.
Conclusion: being a "complete human" in the AI era
The dual mechanism of AI's kill lines reminds us: in the era of the intelligence revolution, humanity must not only learn to use AI tools, but also learn how not to be defined by AI.
The real moat is humanity's unique cognitive framework, emotional experience, and value judgment ability. Only by first becoming a "human" can one evolve together with AI as a human—rather than being alienated by AI, or eliminated by AI.
In this era full of uncertainty, what is needed is not fear and avoidance, but courage and wisdom—bravely embracing the opportunities AI brings, and wisely guarding the essential values of humanity. This, perhaps, is the last line of defense humanity must hold in the AI era, and also the confidence to keep up with the times and not be eliminated.
What do you think of this question? Let's chat in the comments.