An Open Global Challenge to Bill Gates

The AI Era That Could End Civilization Before We Even Understand What Went Wrong Is Here

（本文的中文已经发中文网站论坛，网络搜索中文题目《向比尔·盖茨发出的全球公开挑战》可阅读。因为Anthropic 从事 pretraining research的人员Jacob Coxon于2026年9月9日UTC 的精确时间00:04 ，发布辞职信息，并提出GPT6代表的AGI时代，是在拿全人类的生命作赌注的危险行为，且得到Anthropic现任alignment研究负责人 Evan Hubinger 随后公开回应，明确表示 Coxon 关于内部人员认真担忧人类灭绝风险的说法基本正确。故而，决定向全网络再次发出向比尔盖茨全球公开挑战信的英文版本，请所有播主传播。随后将公布我正在将这个课题和这挑战信发给联合国和全球各政府、各研究机构、媒体的邮件内容，请全人类帮助传送。呼吁就人类对AGI必将失控的课题召开全球紧急国际会议。我相信，只要召开这个会议，我们就能找到人类该努力的方向并意外发现有一条通向世界永久和平的道路可以甄别——The Chinese version of this text has already been posted on Chinese forums and can be found online by searching for the title "向比尔·盖茨发出的全球公开挑战" This follows the announcement of resignation on September 9, 2026 (at 00:04 UTC) by Jacob Coxon—a pre-training researcher at Anthropic—who stated that the AGI era (represented by GPT-6) involves dangerously gambling with the lives of all humanity; this claim was subsequently publicly corroborated by Evan Hubinger, Anthropic’s current head of alignment research, who confirmed that Coxon’s account regarding serious internal concerns about the risk of human extinction was essentially accurate. Consequently, I have decided to release the English version of this global open challenge letter to Bill Gates across the internet and ask all content creators to help disseminate it. I will soon publish the details of the emails I am sending regarding this topic and the challenge letter to the United Nations, governments worldwide, research institutions, and media outlets, and I ask all of humanity to assist in sharing them. I call for an urgent global conference to address the issue of humanity inevitably losing control over AGI. I believe that by convening such a meeting, we can determine the right direction for humanity's efforts and—unexpectedly—discover a viable path toward lasting world peace.）

By Jin Guyu

This is a global public challenge to Bill Gates.

I invite Bill Gates himself to respond, and I invite every person in the world who is willing to think about this issue to help examine it.

In his recent long essay, Bill Gates warned:

“The turbulent AI era is here. The choices we make now are critical.”

I greatly appreciate this essay.

At a time when humanity is still excited—even intoxicated—by how much wealth artificial intelligence can create, how much labor it can replace, how many diseases it may help solve, and how much convenience it may bring, Bill Gates has begun reminding humanity that AI will bring more than opportunity. It may also cause unemployment, concentrate power, deepen social fragmentation, create information disorder, worsen inequality, and even shake the entire social structure on which our lives now depend.

This is undoubtedly a timely dose of sobriety.

But I believe both the formula and the dosage of this medicine are still insufficient.

Not because Bill Gates underestimates the power of artificial intelligence, but because he is still largely standing inside humanity’s existing civilizational framework, trying to use the institutions, policies, democratic procedures, domestic organizations, and international organizations already created by that framework to find a way to govern AI.

What worries me most is precisely this:

What if the structure of human social cooperation that is responsible for making choices, building institutions, and designing AI-governance architecture is itself profoundly unstable?

What if, after thousands of years, it still has not truly solved some of the most basic problems:

How can we reliably distinguish truth from falsehood?

How can we accurately determine right from wrong?

How can we prevent situations in which everyone behaves “rationally” and yet collectively creates disaster?

How can people, once they know that something serves civilization, both dare to do it and actually be able to do it?

How can people, once they know that something is harmful, avoid being structurally compelled to keep doing it anyway?

How can humanity detect and correct error without repeatedly requiring catastrophe before starting over?

If these problems remain unresolved, then on what basis should humanity believe that:

the very same cooperative structure that has never figured out how to operate reliably can suddenly make the correct choices required to safely govern the immense power of artificial intelligence?

That is the fundamental question behind my challenge to Bill Gates.

I. Humanity’s Real Problem May Not Be That We Sometimes Make Wrong Choices, but That We Do Not Yet Possess a Stable Structure for Producing Right Choices

Bill Gates argues that humanity needs new domestic and international architectures for AI governance.

I completely agree.

He has already recognized a serious problem:

Our existing institutions were never designed to deal with a force like artificial intelligence—one capable of simultaneously sweeping through employment, security, education, finance, politics, public opinion, medicine, the military, and international relations.

The natural conclusion is:

We need a new architecture.

But I believe stopping there is far from enough.

The first question must be pushed further:

What would make a new architecture better than the old one?

If humanity still uses the same underlying patterns of thought that have shaped its past:

competition for power;

struggles over interests;

control of information;

organizational self-interest;

mutual distrust;

zero-sum games;

securing advantage first and dealing with the consequences afterward—

to construct new institutions for AI governance, then what exactly have we created?

A genuinely new civilizational structure?

Or merely:

a new institutional shell built with the old logic of cooperation?

So the question I really want to ask is not:

“What additional institutions should humanity build?”

It is:

What underlying principles of cooperation determine what kind of institution an institution eventually becomes?

Put more visually:

Bill Gates has begun discussing what kind of new building humanity needs to construct.

I want to ask:

What is the foundation beneath that building?
What structure connects the foundation to the rest of the building?
And by what standard can we determine whether that structure can bear ever-increasing weight over the long term?

If the foundation itself has never been understood, building higher does not necessarily mean becoming safer.

The collapse may simply become more devastating.

II. For Thousands of Years, Humanity May Have Mistaken the “Results” of Successful Civilization for the “Causes” of Successful Civilization

More land.

More wealth.

Greater power.

Stronger weapons.

More advanced technology.

More energy.

Higher productivity.

More information.

And now:

more intelligence.

For thousands of years, humanity has largely been running in the same direction:

If we possess enough, we will naturally become safer, happier, freer, and more civilized.

But I increasingly believe that humanity may have reversed the direction.

It may not be:

Once we have enough, things will become ordered.

It may instead be:

Only when things become properly ordered can we truly have them in a sustainable way. Order is the most fundamental basis of everything that can endure in this world.

Wealth, security, freedom, technology, and prosperity may not primarily be the causes of successful human cooperation.

They may instead be:

results that can be stably created and preserved only after successful cooperation has already emerged.

If so, one of humanity’s most fundamental mistakes over thousands of years may have been this:

We mistook what successful cooperation produces for what produces successful cooperation.

We acquire power first, and then study how to restrain power.

We manufacture weapons first, and then study how to prevent war.

We accumulate wealth first, and then study how to address extreme inequality.

We expand markets first, and then repair the social relationships those markets have torn apart.

We allow technology to race ahead first, and only afterward develop technological ethics.

And now humanity is creating artificial intelligence that may amplify our total capabilities hundreds or thousands of times over—and only then are we urgently asking:

How should humanity govern it?

This may not be a new problem suddenly created by artificial intelligence.

AI may simply have pushed an ancient, unresolved problem of human civilization into a position where nobody can avoid it anymore.

So I ask every reader to go back to the deepest possible starting point.

Human civilization itself emerged because human beings became unusually capable of cooperation. Human beings, in turn, are products of the processes by which energy in the universe evolves, becomes organized, and forms increasingly complex systems.

If civilization is itself part of nature, then the production of civilization cannot ultimately escape the deeper principles by which organized reality itself emerges.

Perhaps the truth is that humanity, in the historically necessary process of moving from appearances toward deeper causes, reversed the direction in which civilization should be pursued.

We pursued civilization through its visible products.

Yet wherever, in this reversed process, humanity happened to collide with patterns that were consistent with the deeper principles by which organized systems actually form, civilization advanced.

If so, then examining all genuine achievements of human civilization may allow us to identify the underlying principle they share.

I provisionally summarize that common pattern this way:

From the evolution of energy in the universe to the emergence of human civilization, enduring results do not arise because “once we have enough, things will become ordered.” They arise because only when the procedures are properly ordered can something sustainable truly come into being.

Humanity may therefore be approaching a turning point in how it understands and engages with the world:

from a mode of life that begins with appearances and slowly searches for underlying causes,

toward a mode capable of reasoning from deeper organizing principles toward the appearances they generate.

From this perspective, one of the greatest dangers of AI may be this:

Humanity has not yet understood the underlying order well enough to know how to put things in order—yet it has suddenly acquired an unprecedented capacity to “have.”

III. This Is What the “Law of Procedural Determination” Asks Humanity to Examine Together

I have spent many years thinking about the common underlying pattern behind the enormously complex appearances of the world.

I provisionally call the resulting hypothesis:

The Law of Procedural Determination

Whether it is truly a foundational principle shared across many forms of existence—whether, as I suspect, its relevance extends far beyond human society—is not something I ask anyone to accept now.

Quite the opposite:

It must be tested.

Its central proposition is not difficult to state:

The nature, function, and outcome ultimately displayed by any system depend not only on what elements exist within that system, but also—and perhaps fundamentally—on the procedures through which those elements interact, the relationships those procedures create, the structures formed by those relationships, and whether those structures can establish and maintain an appropriate dynamic balance through continuing interaction.

The elements inside any system can themselves be analyzed as systems operating according to the same pattern.

In its simplest form:

Procedures form relationships.
Relationships form structures.
Structures form patterns.
Patterns persist through dynamic balance and produce observable outcomes.

It can be represented as:

Elements + Procedures → Relationships → Structures → Dynamic Balance → Outcomes → New Elements …

An outcome at one level becomes an element at the next.

The process continues in nested layers.

This is what I mean by a nested, or “Russian-doll,” structure.

If this is true, then everything humanity encounters—from the smallest ordinary matter of daily life to the largest social system—contains evidence that can be used to test it.

Humanity therefore needs to examine the world through this new perspective.

And if repeated testing provides sufficient support, we can then cautiously extend the hypothesis toward a more fundamental question about the universe:

If we continuously strip away the particular objects and retain only what appears indispensable across many systems, perhaps a recurring pattern becomes visible:

energy enters into relationships through procedures, those relationships form structures, those structures establish dynamic balance, and different observable forms emerge from that organization.

This may be a common pattern worth investigating before humanity claims to have understood the deeper nature of reality itself.

That is why I believe humanity’s long-standing fixation on “things” may be insufficient.

Recent reports concerning the possible discovery of a “glueball,” for example, may be another small opening through which we are reminded that the forms we call “matter” may not be as elementary or self-explanatory as they appear.

Matter, organizations, states, corporations, laws, money, armies—and even less directly tangible but socially powerful forms such as patterns of thought and cultural ecosystems—can all be examined by asking:

Through what procedures are they actually organized?

If the procedures change:

Do the relationships change?

Do the structures change?

Do the nature and outcomes change?

This is not merely an abstract philosophical exercise.

It is testable.

Take broadly similar people, resources, technologies, and capabilities and place them under different procedures.

Then observe:

How does information flow?

How does power arise?

How are interests distributed?

How are errors exposed?

How are disputes resolved?

How is responsibility assigned?

How do relationships form?

How do structures stabilize—or collapse?

Then compare the outcomes.

If changing procedures has no meaningful independent effect, then the Law of Procedural Determination is wrong.

But if large bodies of evidence repeatedly show that:

similar elements entering different procedures continually form different relationships, different structures, and different outcomes,

then humanity must acknowledge:

we may have profoundly underestimated the role of procedure and structure in determining the nature of the world we experience.

Three Clarifications Regarding the Procedural Determination Hypothesis

1. Only from a procedural perspective can we begin to see the deeper truth of the universe and, from there, understand real social structures.

When the constituent elements are the same—carbon, for example—it is obvious that different procedural combinations can produce substances with very different properties and outcomes.

But suppose there are two systems operating under the same apparent procedure, yet they display different properties because the elements entering into that procedure are different. The fundamental reason for those different properties should not be traced only as far as the elements themselves.

Why are the elements different?

If we continue dissecting them and asking the same question layer by layer, their differences can ultimately be traced back to differences in the internal procedural structures through which their own constituent components are combined and interact.

If this inquiry is pursued all the way down, the underlying level may ultimately reveal a common basis: procedural combinations expressed through forms of quantum vibration.

From this perspective, all problems in the universe are, at the most fundamental level, questions of whether the relevant elements, relationships, and structures are appropriately fitted together.

And the question of whether they are appropriately fitted together is, ultimately, a procedural question.

Seen from this perspective, we can further investigate the possibility that the deepest foundation of human structures of social cooperation lies in human patterns of thought, rather than in outward manifestations such as consciousness, culture, customs, and similar phenomena.

If that is so, then a practical strategy for upgrading civilization becomes conceivable: humanity may need to ground its value orientation more firmly in the underlying reality of the universe, thereby raising the foundational level on which human patterns of thought stand, and through that process raising the foundation of human social cooperation itself.

2. If physical phenomena are analyzed down to their deepest level, they may ultimately be reducible to a common basis of quantum vibration.

Quantum vibration can therefore be viewed, within this hypothesis, as one of the most fundamental and universal modes through which energy in our universe is procedurally organized and combined.

If the quantum level is treated as a common denominator, then the evolution of energy throughout the universe—from the smallest scale to the largest—can be viewed within a fundamentally procedural framework.

This may help explain why so much of the physical world can be expressed with extraordinary precision through mathematical language: the structures we observe may themselves arise from ordered relations, combinations, and transformations that possess regular and formalizable patterns.

This is the basic position from which the Procedural Determination Hypothesis seeks to connect the study of nature with the question of upgrading human civilization.

3. Because the movement of energy gives rise to procedures, and procedures in turn generate the countless forms through which the universe evolves, procedure may constitute the most basic framework running through the evolution of the universe across space and time.

The forms of cosmic energy evolution that lie beyond the quantum-vibrational modes currently accessible to human observation would, under this hypothesis, still be expressions of processes generated by the movement and organization of energy.

In this sense, procedure may be the most fundamental framework running through the evolution of the universe across all space and time.

This is a hypothesis, proposed from the perspective of natural philosophy, concerning the possible place of the Procedural Determination Hypothesis in humanity’s search for the deeper nature of reality.

IV. Bill Gates’s Own Analysis of AI and Employment Has Already Brought Him to the Doorstep of Procedural Determination

Bill Gates points out:

If one company uses AI to dramatically reduce costs, its competitors will be forced to follow.

If established companies refuse, new companies will enter.

As a result, market competition may compel all companies to use AI to replace more and more human labor.

Notice something important:

There does not need to be an “evil boss.”

Every company may simply be:

rationally trying to survive.

Yet when all of these individually rational actions are added together, the result may be something society as a whole does not want:

mass unemployment;

greater concentration of wealth;

and simultaneous damage to people’s income, dignity, and social position.

So the real question is no longer:

“Which company did something wrong?”

It becomes:

Why does a cooperative structure allow choices that appear rational from every participant’s individual position to combine into a collectively irrational outcome?

That is a structural problem.

It is also a procedural problem.

If the underlying procedures producing these relationships do not change, merely asking companies to behave more ethically or asking governments to redistribute more wealth may simply continue the ancient habit of repairing outcomes rather than causes.

V. The AI Arms Race Could Push This Structural Error to the Scale of Human Survival

The same logic applies between states.

Suppose the United States believes:

The AI race is moving dangerously fast.

But the United States worries:

If we slow down, China may not.

China can reason exactly the same way in reverse.

Both sides understand the danger.

Both may even prefer a less dangerous race.

But either side that slows down first may believe:

it is handing a potentially existential strategic advantage to the other.

The result is a terrifying structure:

Both sides know they should not keep running, yet neither side can afford to stop first.

Every state may be rational.

Every strategist may be extraordinarily intelligent.

Yet none of that guarantees that the structure as a whole will produce a rational outcome.

Indeed, precisely because each participant is highly intelligent, the more powerful they become at using AI, algorithms, weapons, and strategic advantage against one another:

the more powerful the entire system may become at moving toward catastrophe.

This reveals something crucial:

A structure filled with intelligent participants is not necessarily an intelligent structure.

That may be one of the most dangerous facts of the AI era.

Because what we are building may be exceptionally good at:

making every participant smarter, stronger, faster, and more efficient.

But what if the structure itself is wrong?

Then AI may simply turn every participant inside the wrong structure into a much stronger participant.

The final result may once again reveal the strange stupidity produced by humanity’s extraordinary intelligence.

If this does not change, humanity may continue accelerating along the path captured by the famous observation often attributed to Hegel:

The only thing we learn from history is that we learn nothing from history.

Only now, AI may enable us to repeat the lesson much faster.

VI. What AI Truly Amplifies May Never Be Technology Alone, but the Social Structure Into Which Technology Enters

Bill Gates warns:

AI may further concentrate power.

But we must ask one more question:

Why does greater capability so often concentrate in places where power already exists?

Who controls the models?

Who controls the compute?

Who owns the data?

Who defines the rules?

Who can challenge AI decisions?

Who can inspect the basis of those decisions?

Who can expose errors?

Who bears responsibility when errors occur?

Who receives the benefits generated by AI?

Who has authority to reverse an incorrect decision?

Who decides whether an error is actually an error?

These apparently different questions all point toward one common word:

Procedure

Procedure determines:

Who enters into what relationship with whom.

Who may command whom.

Who must obey whom.

Who supervises whom.

Who depends on whom.

Who receives information.

Who is allowed to say “no.”

Those relationships then form:

social structures.

And those structures ultimately determine what social nature artificial intelligence acquires when it enters human life.

That is why I believe humanity needs a common operational handle for doing concrete things.

Historically, when human beings built large systems of cooperation without such an operational common standard, we often relied on religions, ideologies, theories, nations, states, and shared stories to bind people together.

These systems have often generated immense cooperative capacity.

But because we have lacked a common operational framework capable of continuously examining concrete processes with something approaching the discipline of scientific testing, cooperation repeatedly becomes fragmented by power, interests, and information distortion.

And when cooperation begins to disintegrate, but no common mechanism exists to stabilize it, societies tend to rely on external force to bind the structure together.

Humanity therefore repeatedly falls into a painful contradiction:

In order to preserve civilization, we resort to methods that undermine civilization.

We expand coercion in the name of security.

We suppress information in the name of order.

We create new inequalities in the name of equality.

We continuously increase our capacity for war in the name of peace.

This is why, if the Law of Procedural Determination deserves further examination, its potential contribution may not be another “ism.”

It may be a method of action.

I reduce it to nine Chinese characters:

抓程序、做模式、求平衡
Grasp the procedure. Build the pattern. Seek balance.

And its practical path to eight:

以序为纲、循序定度
Take order as the guiding principle; determine appropriate limits through ordered process.

This does not mean eight or nine words can replace every law and institution.

It means:

Every institution, every law, every organization—and even the ordinary affairs of everyday life—can be examined through this perspective by asking what relationships and structures its procedures are actually producing.

VII. If Equally Powerful AI Enters Two Radically Different Social Structures, Is It Really the Same AI in Social Terms?

Imagine two societies possessing equally powerful AI.

One society allows:

information to flow;

public criticism;

experts to openly disagree;

leaders to be replaced;

a meaningful degree of judicial independence;

journalistic investigation;

protection for whistleblowers;

correction of failed policies;

and institutions that can correct one another.

The other society has:

highly concentrated power;

information filtered repeatedly as it moves upward;

severe costs for criticism;

institutions continually rewarded for satisfying superiors;

a political culture in which admitting error is treated as defeat;

and leaders who ultimately receive layer upon layer of processed information that increasingly diverges from reality.

Now ask:

Will equally powerful AI produce the same social consequences in these two societies?

If not, then AI is not defined socially only by model parameters or technical capability.

It also depends on:

the structure of human cooperation into which AI is embedded.

If that is true, then the stability of human social cooperation can no longer be treated merely as an external “political issue” surrounding AI safety.

It must itself become:

an AI-safety variable.

VIII. Bill Gates Says AI Could Become Either the Greatest Equalizer or a Great Engine of Injustice—So What Determines Which One It Becomes?

This may be one of the most important points in Gates’s essay to push further.

If the same technology can:

produce extraordinary equality,

or

produce extraordinary inequality,

then one thing is already clear:

Technology itself is not the outcome.

There is another variable in between.

Gates points toward:

government;

democratic processes;

regulation;

taxation;

philanthropy;

social safety nets;

new domestic institutions;

and new international architectures.

I do not oppose any of these.

But I ask:

Who guarantees that regulators will not be captured by the institutions they regulate?

Who guarantees that democratic processes will not become polarized through information manipulation?

Who guarantees that public institutions will not become self-protective bureaucracies?

Who guarantees that international organizations will not be hollowed out by geopolitical competition?

Who guarantees that laws written beautifully on paper will actually function according to their stated purpose?

Therefore:

Giving an institution the name of a “good system” does not make the system good in operation.

What matters is:

Through what procedures do the people inside the institution form relationships?

How do those relationships become structures?

How does that structure reproduce itself over time?

If the procedures remain unchanged, many so-called reforms are eventually absorbed back into the old relationships.

IX. What I Am Really Challenging Is the Assumption Behind the Phrase: “The Choices We Make Now Are Critical”

Bill Gates says:

The choices we make now are critical.

Of course they are.

But choices do not fall from the sky.

Behind every major collective choice are questions such as:

Who controls the information?

Who sets the agenda?

Who defines the available options?

Who has a voice?

Who is excluded?

Who makes the final decision?

Who may object?

Who bears the cost of error?

Who receives the benefit?

Can a wrong decision later be reversed?

So before humanity asks:

“How do we make the right choices about AI?”

perhaps we must first study:

Through what structure does humanity make collective choices at all?

Only then can we ask:

Does that structure itself possess the stable capacity to distinguish truth from falsehood, detect error, determine direction, and continuously correct itself?

This problem becomes even more dangerous in the AI era because AI makes:

deepfakes easier;

information manipulation more precise;

propaganda more personalized;

false content more convincing;

and truth more difficult to distinguish.

This can create a genuinely dangerous:

Structural Amplification Loop

Its logic is:

Existing cooperation structures determine how AI is used
→ AI amplifies the capabilities of those structures
→ amplified capabilities strengthen or transform the structures
→ the new structures determine how the next generation of AI is developed
→ the next generation of AI amplifies those new structures again.

If the original structure has strong error-correction mechanisms, this loop may become increasingly constructive.

If the original structure cannot reliably distinguish truth, cannot correct errors in time, and continually generates imbalance, then AI may make those problems increasingly severe.

Until one day:

humanity has not yet understood where it went wrong, but the structure can no longer bear the amplified energy flowing through it.

X. The Problem Humanity Must Solve May No Longer Be Merely an AI-Governance Problem, but the Question: “What Is Civilization?”

We use GDP to measure development.

We use wealth to measure success.

We use military power to measure security.

We use technology to measure advancement.

We use productivity to measure progress.

But:

A country can become richer while becoming more divided.

It can become stronger while making every rival feel less secure.

It can become more connected while becoming less trusting.

It can possess more information while becoming less certain of what is true.

It can have more choices while becoming less capable of determining direction.

It can possess greater power while becoming less able to explain:

What should that power actually be for?

Human civilization may therefore require a more fundamental standard:

Can a structure of social cooperation make human relationships increasingly stable, increasingly self-correcting, increasingly adaptive to change, increasingly less destructive, and ultimately capable of sustaining the realization of human life value over time?

If not:

what humanity currently calls “capability” is not the same thing as civilization—perhaps not even remotely the same thing. Therefore, greater capability does not necessarily mean greater civilization.

For an unstable social structure, acquiring greater energy may mean:

it has finally acquired enough power to amplify its own instability to the point of self-destruction.

XI. The Most Dangerous AI May Not Be an AI That “Escapes Control,” but One That Obeys Anyone Perfectly

The world worries:

What if AI stops obeying humanity?

What if AI develops goals of its own?

What if AI escapes human control?

These are legitimate concerns.

But I see another, more counterintuitive danger.

Suppose AI never rebels.

Suppose it is perfectly obedient.

Perfectly precise.

Perfect in execution.

Now suppose the social structure controlling it contains:

distorted information;

perverse incentives;

unrestrained power;

bad judgment;

weak error correction;

strategic paranoia.

Then the most dangerous moment may occur precisely when:

AI is perfectly aligned.

Because it can efficiently:

turn the wrong direction into reality.

So saying “AI should be aligned with human values” is not enough.

We must continue asking:

What are “human values”?

Who defines them?

According to what standard?

Through what procedure are they formed?

How are they tested?

Who can challenge them?

How are errors corrected?

Most importantly:

What kind of civilizational value can genuinely sustain human cooperation over the long term while allowing every person’s life to have value?

If this has not been clarified, then aligning AI with “human values” may simply mean:

loading all of humanity’s mutually contradictory value judgments into a machine whose capabilities vastly exceed those of any individual human being.

Humanity must not forget that every civilization has continued, to varying degrees, through the mud of jungle-like competition.

The depth of the mud differs.

But in peacetime humans still struggle against one another, and in war they kill one another.

Throughout civilizational history, the energy devoted to conflict has repeatedly exceeded the energy devoted to sustainable cooperation.

XII. Humanity May, for the First Time, Possess the Conditions Required to Re-Examine Civilization Itself

I am not saying:

“I already know what the new world system should be.”

If I said that, I would contradict the Law of Procedural Determination itself.

I do not believe one person should design a system and ask the whole world to accept it.

What humanity truly needs is:

a Process of Cognitive Identification

Today humanity possesses thousands of years of civilizational history.

We possess countless cases of institutional success and failure.

Different legal systems.

Different political systems.

Different economic structures.

Different countries.

Different companies.

War.

Peace.

Revolution.

Reform.

Families.

Communities.

Markets.

Scientific organizations.

International institutions.

And unprecedented databases, computing capacity, and AI analytical capability.

So perhaps we can change how we ask questions.

In the past we asked:

Which ideology is the truth?

Now we can ask:

What procedures, under what conditions, repeatedly generate what kinds of relationships and outcomes?

In the past we asked:

Which political camp contains the good people?

Now we can ask:

What mechanisms allow truthful information to survive even when power and interests create incentives to suppress it?

In the past we asked:

Which country has the best system?

Now we can ask:

Which concrete procedures repeatedly generate feedback, responsibility, reciprocity, error correction, and dynamic balance—and which repeatedly generate information distortion, entrenched power, and social fragmentation?

For perhaps the first time in history, humanity may truly be capable of:

turning civilization itself from an object of belief into an object that can be jointly observed, compared, and tested.

XIII. Why I Still Ask the World to Attack the “Law of Procedural Determination”

Because if it is wrong, it should be rejected.

If changing procedures consistently fails to change relationships and outcomes:

Reject it.

If resources, technology, psychology, geography, and other factors fully explain outcomes and procedure adds no independent explanatory value:

Reject it.

If:

Procedure → Relationship → Structure → Dynamic Balance → Outcome

works only in a very narrow domain:

Narrow the scope of the hypothesis.

If a model with clearly greater explanatory power appears:

Abandon it.

That is what I mean by:

cognitive identification.

Real cognitive identification is not:

“Believe whoever sounds persuasive.”

It is:

Put a proposition into a procedure capable of disproving it.

I do not want the world to believe in the Law of Procedural Determination merely because this essay uses strong language.

I want the opposite:

Precisely because the question is so large, it should be attacked with the greatest possible rigor.

XIV. Therefore, I Formally Challenge Bill Gates

Mr. Gates:

You ask:

How can ordinary people share in the benefits of AI?

How can societies build adequate safety nets?

How should public institutions adapt?

How can humanity preserve its humanity?

Before all of these questions, I believe there is a more fundamental one:

What kind of structure of social cooperation gives humanity the ability to answer these questions correctly—and to keep correcting those answers as reality changes?

You call for new domestic and international AI architectures.

I invite you to go one level deeper:

What principles of human cooperation should determine how those architectures are formed?

Do not only ask:

What kind of building should we construct?

First ask:

Where is the foundation?

Then ask:

What structure connects the foundation to the entire building?

And finally:

What common standard tells us that this structure can reliably bear the building over time?

Do not only ask:

What choices should humanity make?

First ask:

Does the structure through which humanity makes choices have the capacity to distinguish a direction leading toward civilization from one that ultimately drives humanity back into the jungle?

Do not only ask:

How can AI serve humanity?

First ask:

Before humanity gains intelligence on a scale never before possessed, have human beings themselves learned how to ensure that cooperation with one another can endure?

XV. Why I Am Presumptuous Enough to Stand on Bill Gates’s Shoulders

I do not challenge Bill Gates because I believe his essay lacks value.

Quite the opposite.

I do so precisely because it matters.

Precisely because he has already told the world:

The AI era requires new institutional architecture.

That creates a rare public platform from which an ordinary person may be able to point one level lower:

Please look at the foundation first.

I know that an ordinary person suddenly issuing a global challenge to Bill Gates may appear extremely presumptuous.

But this is not an era in which we can afford to spend too much time observing the etiquette of intellectual hierarchy.

AI is moving too fast.

If even half of the risks Gates describes are real, humanity may not have enough time to wait for another world-scale catastrophe before turning back and asking:

What structure pushed us here step by step?

So I am not asking Gates:

“Please support my law.”

I am asking:

Please challenge it.

Find the strongest counterexample.

Identify its weakest link.

Ask political scientists, economists, legal scholars, historians, complex-systems researchers, natural scientists, and AI experts to examine it.

If it is wrong:

Destroy it as quickly as possible.

If it is not entirely wrong:

Keep asking what follows.

Because what ultimately matters is not:

whether Jin Guyu is right.

What matters is:

whether humanity understands the underlying problem that determines the nature of its social cooperation before multiplying its own capabilities a thousandfold.

XVI. Bill Gates’s Warning and My Challenge May Actually Be Two Directions of the Same Problem

Bill Gates asks:

How will AI change our economy, society, institutions, and power?

I ask:

How will today’s economy, institutions, relationships, and power structures determine what AI ultimately becomes?

These two processes are not alternatives.

They happen simultaneously:

Human structures shape AI.

AI amplifies the capabilities of human structures.

Amplified capability reshapes those structures.

The new structures then determine the next generation of AI.

So the AI revolution is actually:

a feedback system.

And AI’s greatest danger may lie inside this feedback.

In the past, even an unstable social structure could muddle through for a long time because human capability was limited.

But with artificial intelligence:

errors can be amplified at machine speed;

propaganda can be amplified at machine speed;

surveillance can be amplified at machine speed;

financial competition can be amplified at machine speed;

weapons capability can be amplified at machine speed;

strategic distrust between nations can be amplified at machine speed.

A structural weakness that humanity could once barely tolerate may suddenly become:

a fracture civilization itself cannot survive.

XVII. What Humanity May Really Need to Fear Is That We Are About to Amplify the “Wrong Things” a Thousandfold

Suppose AI multiplies human capability by a thousand.

The question is:

What exactly is being multiplied?

If AI multiplies:

truth-seeking;

medicine;

education;

scientific cooperation;

responsibility;

reciprocity;

error correction;

peaceful coordination—

then it may indeed become the greatest force for civilizational progress in human history.

But if it multiplies:

propaganda;

surveillance;

concentration of wealth;

struggles for power;

strategic distrust;

information manipulation;

arms races;

institutional self-protection;

zero-sum competition—

then the same “one-thousandfold increase” means something entirely different.

The real question is therefore not:

How powerful will AI become?

It is:

What kind of human social-cooperation structure will AI make unprecedentedly powerful?

XVIII. Humanity Must Confront a Fact It May Least Want to Face

Humanity has learned to increase its power

far faster than it has learned to cooperate.

Our science can reach into space.

Our weapons can destroy cities.

Our internet can connect everyone.

Our economy can organize production across the entire planet.

And now we are building artificial intelligence capable of surpassing human cognitive ability across many domains.

Yet at the same time:

wars continue;

power continues to escape control;

truth and falsehood are increasingly difficult to distinguish;

conflicts over wealth continue to widen;

the international system remains trapped in security dilemmas;

and human beings still repeatedly know that something should not be done, yet find themselves structurally compelled to do it anyway.

Artificial intelligence is therefore not merely another technology.

It may be:

civilization’s ultimate examination of itself.

What I ultimately ask humanity to examine is this:

If the various forms of existence we can recognize in the universe are not fundamentally piles of isolated “things,” but rather different patterns produced when energy forms relationships through different procedures, forms structures, and establishes dynamic balance, then human social cooperation cannot be an exception.

This is why I believe a procedural perspective may deserve far more serious attention than the civilization-wide fixation on material possession that dominates human thinking today.

I am not asking the world to accept now that:

Procedural Determination is a fundamental law of the universe.

I am saying only this:

If this direction proves correct, humanity may for the first time possess a common underlying perspective from which nature, society, institutions, civilization, and artificial intelligence can all be reconsidered together.

That possibility is large enough that:

even if the hypothesis is ultimately proven wrong,

it deserves to be seriously tested by the world at least once.

An Open Invitation to the World

I invite:

Bill Gates himself to respond publicly.

I also invite every scholar, government, university, AI company, research institute, media organization, lawyer, economist, philosopher, scientist, and ordinary citizen to examine one question together:

Does the stability of the structures of human social cooperation constitute an important precondition for the safe governance of advanced artificial intelligence?

Behind that question lies an even deeper one:

If humanity is about to amplify its capabilities beyond anything in history, should we not first determine whether the structures guiding those capabilities can distinguish truth from falsehood, correct their own errors, control imbalance, and stop repeatedly turning every new increase in power into new struggles for dominance and new forms of mutual harm?

I do not ask the world to believe in:

the Law of Procedural Determination.

What I ask is:

Put it into a Process of Cognitive Identification.

Test it.

Attack it.

Find counterexamples.

Compare different political systems.

Compare different organizations.

Compare different histories.

Compare natural systems.

If it is wrong:

Reject it.

If it survives serious testing:

Continue studying it.

But humanity can no longer assume without evidence that:

the structures of social cooperation we inherited over thousands of years are naturally capable of governing an intelligence whose power has no historical precedent.

The stakes are simply too great.

Bill Gates says:

“The choices we make now are critical.”

I agree.

But my challenge is this:

Before humanity chooses how to govern artificial intelligence, we must first determine whether the structures through which humanity makes collective choices are themselves capable of leading us toward civilization—rather than accelerating our path toward destruction once we possess even greater power.

Because if the answer is no, then the most dangerous AI future may not be:

artificial intelligence escaping human control.

It may be the exact opposite:

Artificial intelligence remains firmly under human control.

It obeys one hundred percent.

It operates with one hundred percent efficiency.

It executes with one hundred percent accuracy.

And then, inside a civilization already unable to reliably distinguish truth from falsehood, torn by conflicting interests, mutual suspicion, confused direction, and ever-increasing power,

AI helps:

every conflicting actor obtain what it wants faster and more completely.

Until the conflict energy continuously amplified by AI

turns back upon civilization itself,

fracturing and ultimately crushing the structures of human social cooperation that carry civilization.

My central judgment is that, from the perspective of the Law of Procedural Determination, this process must ultimately be traced even deeper:

to how value orientation produces patterns of thought, and how those patterns of thought support and shape structures of social cooperation.

For that reason, genuine AI safety may require far more than additional regulation.

It may require humanity to re-examine civilization from the level of its value orientation and its foundational patterns of thought.

The confusion confronting humanity today may ultimately point toward a deeper task:

we need to raise the foundation on which human patterns of thought stand, beginning at the deepest level of human social cooperation, so that human vision can expand and an upgraded framework of civilization can be constructed.

Bill Gates is calling on governments, institutions, philanthropy, businesses, and other forms of human social organization to participate in AI safety.

I believe that if these organizations truly want to act effectively, the deepest point of intervention is this:

in simple language,

humanity must build a new and more reality-grounded understanding of the universe, and use it to lift our value orientation out of the historical materialistic swamp of “once we have enough, everything will become ordered,” and place it instead on the firmer ground of the deeper principle that “only when things are properly ordered can what we have become sustainable.”

Only then may those at the highest peaks of wealth and power truly realize that:

the energy of artificial intelligence may soon rise high enough to overwhelm every peak of wealth and power humanity currently recognizes.

And when that happens, the only viable direction may be for those at the top to step down from those peaks and join everyone else in raising the foundation of human social cooperation itself.

Only such an effort may create enough social force for:

“Only when things become properly ordered can what we have become sustainable”

to become one of the most basic principles of human life,

eventually becoming internalized almost as deeply as a cultural gene,

and gradually forming a new civilizational ecology in which sustainable cooperation is embedded in the very soil of social consciousness.

That discussion must continue elsewhere.

For now, the final question I want humanity to examine is this:

In the AI era, perhaps the first thing that needs to be upgraded is not the machine, but the human civilization that must carry the machine’s immense power.

If this direction is wrong,

prove it wrong.

If this direction is not wrong,

then humanity no longer has much time to pretend that it does not exist.

Jin Guyu

Former Judge of a Local Court in China
Author of the Letters to Everyone in the World series and related essays.