一名中國前高級法官公開致世界AI領軍人物信之五：Kevin Kelly——人類是否忽視而遺漏了支撐文明的底層規律？

One of a Series of Open Letters from a Former Senior Chinese Judge to the World’s AI Leaders, No. 5: Kevin Kelly — Has Humanity Overlooked a Foundational Principle That Sustains Civilization?

（Special Note:

The phrase “former Senior Judge” appears in the title only to give this letter a slightly better chance of being noticed. It is not intended to make readers focus on who I am, nor on the particular way in which I express what I wish to say.

I am simply a very ordinary old man. What I truly wish to discuss is also a principle that appears extraordinarily ordinary — so ordinary that no one can do anything without relying on it. Precisely for that reason, it is like air: because human beings are so familiar with it, it has long been overlooked.

What I truly hope all humanity will examine together is this:

Does such an “ordinary” foundational principle in fact determine the success or failure of human behavior, social cooperation, and even the rise and decline of civilizations? Could it be a basic sustaining principle of civilization that humanity has never fully recognized?

Therefore, who is saying it is not important. Whether what is being said is true is what deserves everyone’s attention.

Since 2014, I have continuously written to governments, contacted the media, social organizations, scholars, and people from many different fields, and gradually published the series Letters to Everyone in the World together with related articles. Today, in deciding to make public some of my letters to leading figures in global AI, I still have only one purpose:

to place this question before all humanity for collective examination.

I believe that when a question may concern the direction of human civilization as a whole, governments and international organizations have a responsibility to study it, the media have an obligation to help the public understand it, and every person has the right to participate in judging it.

What humanity faces today is no longer merely a “crossroads” in the conventional historical sense. If the underlying way in which civilization itself operates contains a fundamental problem, then moving forward, backward, left, or right may all amount only to repeating the old pattern — and this time, the repetition may bring humanity closer than ever before to an accelerating path toward its own destruction.

What truly needs to change is an upward transformation. Perhaps the first thing that must be elevated is the human mode of thinking through which we observe the world and judge value.

The core proposition I put forward is:

Procedures form relationships; relationships form structures; structures generate dynamic balance through feedback; and this ultimately appears as the properties and outcomes of a system.

From this, I summarize the most basic practical standard for civilization as:

grasp the procedure, build the pattern, seek the balance.

I believe the greatest difference between this and many grand theories of civilization in the past is that it does not first require people to accept a particular religion, ideology, or concrete value objective. Instead, it attempts to provide a common practical principle that everyone can understand, that can be applied to real action, and that can be tested against reality.

If this direction is wrong, I hope humanity will help prove it wrong.

If it truly touches upon some foundational principle, then humanity may be able to use it to rebuild and strengthen the underlying foundations of social cooperation, allowing civilization gradually to move away from its long-repeated cycles of mutual struggle, mutual harm, and war, and toward stable, harmonious, and sustainable cooperation.

That is the only reason I am making these letters public.

I do not hope the world will focus on the person who wrote these letters. I hope the world will seriously examine the questions these letters are actually asking.

Finally, a note: some websites impose word-count or character limits, so certain versions of these letters found online may be incomplete. If a text appears incomplete, you may search for the full version using the Chinese or English title of the letter, or contact the publicly listed email addresses of AI companies associated with the person to whom the letter is addressed and ask for assistance in locating or forwarding the complete version.

——特別題註：

標題中寫入“前高級法官”，只是為了讓這封信多獲得一點被看見的機會，並不是希望讀者關注我是誰，以及在用什麼方式表述所說之事。

我只是一個極其普通的老人。我真正想講的，也是一條看起來極其普通的規律——普通到任何人做任何事都離不開它，也因此這規律就像空氣一樣，因為人過於熟悉而長期被忽略。

我真正希望全人類共同鑑別的是：

這樣一個“普通”的底層規律，是否事實上決定着人的行為、社會合作乃至文明興衰的成敗？它是否正是人類長期沒有真正認清的文明底層支撐規律？

所以，誰在說並不重要，說的事情是否真實，才值得所有人關注。

自2014年以來，我持續向政府寫信，也不斷聯繫媒體、社會組織、學者和各領域人士，並陸續發表《致全球每一個人的信》及相關文章。今天決定公開部分致世界AI領軍人物的信，仍然只有一個目的：

把這個問題交給全人類共同甄別。

因為我認為，對這樣一個可能關係整個人類文明方向的問題，政府和國際組織有責任研究，媒體有義務幫助公眾了解，每一個人也都有參與判斷的權利。

今天人類面對的，已經不僅是通常歷史意義上的“十字路口選擇”。如果原有文明運行方式本身存在底層問題，那麼前後左右可能都只是舊模式的重複，且這次可能是最接近不斷加速毀滅全人類的一次。

真正需要改變的是向上抬升，也許首先是人類觀察世界和判斷價值的思維模式。

我提出的核心判斷是：

程序形成關係，關係形成結構，結構通過反饋形成動態平衡，並最終表現為系統的性質與結果。

由此，我把文明最基本的做事標準概括為：

抓程序、做模式、求平衡。

我認為，這與過去許多宏大文明理論最大的區別在於：它並不首先要求人們接受某一種宗教、主義或具體價值目標，而試圖提供一個人人都能夠理解、能夠用於具體做事、也能夠接受事實檢驗的共同抓手。

如果這個方向是錯的，希望全人類幫助證明它錯。

如果它確實觸及了某種底層規律，那麼人類也許可以由此重新夯實社會合作結構的基礎，使文明從長期反覆的互斗、互害和戰爭循環中，逐步走向穩定、和諧、可持續的合作。

這也是我公開這些信件的唯一原因。

不是希望全世界關注寫信的人，而是希望全世界認真看看這些信究竟提出了什麼問題。

最後說明：部分網站存在字數或字符限制，因此網上看到的某些信件可能不是全文。如內容不完整，可以按照信件的中文或英文標題搜索全文，也可以通過與致信對象相關AI企業公開郵箱的聯繫方式，請其協助查找或轉交完整版本。）

To Kevin Kelly: When Artificial Intelligence Enters a “Jungle Civilization,” Does It Strengthen Civilization — or Humanity’s Capacity for Mutual Struggle and Harm?

Dear Mr. Kevin Kelly,

Greetings.

My  pen name is Jin Guyu. I am an elderly Chinese citizen and formerly served as a Senior Judge in a local court in China.

I follow your thinking, and one important reason is that you do not view technology merely as a tool in human hands. Rather, you place technology within a much longer process of evolution, observing how it expands possibilities, changes humanity, and changes civilization itself.

For that reason, I especially wish to ask you a question.

It can be expressed very simply:

When artificial intelligence enters a human civilization that still operates to a great extent according to the logic of the jungle, what does it strengthen first: humanity’s capacity to create civilization, or humanity’s capacity to struggle against and harm one another?

If the answer is mainly the latter, what happens next?

Will humanity be completely destroyed by some accident, misjudgment, escalation of competition, or runaway feedback — or will it be forced to evolve along an entirely different civilizational path?

That is the question I truly wish to ask you in this letter.

By “jungle civilization,” I do not mean that humanity today has no civilization.

Quite the opposite.

Humanity has created astonishing achievements in science, law, institutions, art, ethics, and technology.

But if we examine the underlying structure of social cooperation, we still seem to rely heavily on some very ancient mechanisms to maintain security and order:

competition;
deterrence;
balance of power;
exchange of interests;
control;
defense;
mutual suspicion;
and preventing others from harming us by possessing an even greater capacity to harm them.

States do this.

Corporations do this.

Political groups do this.

Many interpersonal relationships do this as well.

This gives rise to a strange paradox of civilization:

For the sake of peace, we continually increase our capacity for war.
For the sake of security, we continually increase our capacity for mutual deterrence.
In order not to be controlled, we seek greater power to control others.
In order to prevent ourselves from being harmed, we first acquire a greater capacity to inflict harm.

In the past, although this structure was dangerous, humanity may still have had time to correct itself after one disaster after another.

After wars, humanity created new rules.

After crises, humanity redesigned institutions.

After mistakes, humanity slowly learned.

But AI may change the timescale.

AI is not entering an empty human world.

It is entering our already existing:

relations among states;

military systems;

commercial competition;

financial systems;

political struggles;

information systems;

ideological conflicts;

individual desires;

and existing structures of human cooperation and confrontation.

So perhaps the real question is not:

“Is AI itself good or evil?”

But rather:

“Once AI enters today’s human relationship structures, which forces will call upon it first, and which structures will it amplify first?”

If AI increases scientific research capacity a hundredfold,

it can certainly help cure disease.

But it can also increase weapons research capacity a hundredfold.

If AI increases information capacity a thousandfold,

it can certainly help humanity gain knowledge.

But it can also increase deception, manipulation, surveillance, and cognitive warfare a thousandfold.

If AI greatly improves decision-making and organizational efficiency,

it can help humanity cooperate,

but it can also allow already competing states, corporations, and organizations to compete far more effectively.

Therefore, what I truly worry about is not:

AI suddenly becoming bad.

It is:

AI very faithfully and efficiently helping us continue doing things according to the structures we already have today.

If the underlying structure itself contains a problem, then an increasingly powerful AI may first make that problem increasingly powerful.

This leads me to a very simple question:

What is the relationship between the speed of technological progress and the speed of progress in human social cooperation structures?

If technological capability grows exponentially,

while the human ways of handling interests, power, security, responsibility, and mutual trust can still only be corrected slowly through war, crisis, and catastrophe,

will there emerge an ever-widening “civilizational speed gap” between the two?

Technology becomes stronger every day.

But humanity’s ways of cooperating may still remain trapped in structures that have repeated themselves for centuries or even millennia.

If this speed gap continues to widen, then what AI first brings humanity may not be an entirely new civilization,

but rather:

an increasingly powerful engine installed in an old civilization.

The question then becomes:

Where will an old direction, combined with immense new power, take us?

After thinking about this problem for many years, I developed a hypothesis that still needs to be tested.

I call it:

the Procedural Determination Hypothesis.

Its simplest expression is:

Procedures form relationships; relationships form structures; structures generate dynamic balance or imbalance through continuous feedback; and this ultimately appears as the properties and outcomes of a system.

It can also be written as:

procedure → relationship → structure → dynamic feedback → system properties and outcomes.

I do not know whether this hypothesis will ultimately prove valid.

I would rather have scientists search for counterexamples, test it, and even prove it wrong.

But if there is some truth in this direction, then perhaps when we examine AI, we should not ask only:

How intelligent is the model?

What values does the model possess?

Does the model obey humans?

We must also ask:

What procedures does it enter?
With whom does it form relationships?
What permissions does it possess?
What interests drive it?
What feedback does it receive?
And what kind of social structure does it ultimately enter?

Because the same technology, once placed within different relationship structures, can produce entirely different civilizational outcomes.

So I would like to place the question before you more directly.

You have spent much of your work thinking about technological evolution.

From an evolutionary perspective, which of the following paths do you think is more likely?

Path One: AI Strengthens Civilizational Capacity

AI improves knowledge, creativity, production, medicine, education, and social coordination.

Humanity gradually uses new intelligence to reduce resource conflict, improve institutions, increase transparency, and expand cooperation.

AI therefore becomes a force driving civilizational advancement.

This is the future we most hope to see.

Path Two: AI First Strengthens the Capacity for Mutual Struggle and Harm

States use AI against states.

Corporations use AI to defeat corporations.

Political forces use AI to compete for power.

Individuals use AI to compete for interests.

From each side’s local perspective, this may be rational:

“Others are becoming stronger, so I cannot afford not to become stronger.”

The result, however, is an increasingly powerful positive feedback loop:

You strengthen — I strengthen.

I strengthen — you feel less secure.

You feel less secure — you strengthen further.

In the end, everyone becomes more powerful than before,

yet no one becomes safer.

If this process continues,

superintelligence may never need to “betray humanity.”

Human beings themselves may use superintelligence to increase their capacity for mutual struggle and harm to a level civilization can no longer withstand.

If the second path really emerges,

then the third question I most want to ask you is:

How will humanity evolve from there?

One possibility is:

destruction.

A seemingly minor error;

a military misjudgment;

an unexpected action by an AI agent;

a chain reaction involving finance, energy, nuclear weapons, biotechnology, or cyber systems,

amplified by extreme speed and enormous capability,

may eventually cross a threshold from which humanity can no longer return.

That may not be anyone’s plan.

There may not even be an “enemy” in the ordinary sense.

Every participant may simply be doing something “rational” according to a local objective,

yet together they may produce a catastrophic global outcome.

But is there another possibility?

Namely:

Could immense technological pressure force the structure of human social cooperation itself to undergo a fundamental evolution?

Could humanity, perhaps for the first time, be forced to recognize that the real issue is not merely:

who possesses stronger technology;

who controls more resources;

who wins the competition;

but rather:

By what procedures are our relationships organized?

In other words,

could technological evolution eventually force civilization itself to undergo an evolution of social cooperation structures?

If, in the past:

scarcity of resources pushed humans toward cooperation;

war pushed the formation of states;

trade pushed the creation of international rules;

nuclear weapons pushed humanity toward certain new strategic constraints,

then could artificial intelligence become an even stronger evolutionary pressure:

forcing humanity to move gradually from a civilization that maintains security through the capacity for mutual harm toward one that maintains security through stable structures of cooperation?

If so,

then perhaps the greatest revolution ultimately driven by AI

will not be the machine intelligence revolution,

but rather:

a revolution in how human beings cooperate with one another.

For this reason, I have proposed another still-immature concept:

Human Social Cooperation Engineering.

It is not an attempt to turn human beings into machines.

Rather, I simply wish to ask:

If humanity can already systematically study:

why bridges collapse;

why aircraft lose control;

why networks fail;

why AI systems produce unintended behavior,

then can we also study more systematically:

why social cooperation becomes unstable?

Can we study:

What kinds of procedures tend to generate trust?

What kinds of relationships inevitably amplify suspicion?

What kinds of power structures continually reinforce themselves?

What kinds of feedback push all participants toward outcomes that none of them actually desire?

How can each participant protect legitimate interests without destroying the conditions under which the entire system can continue to cooperate?

If these questions can be observed, compared, modeled, tested, and falsified,

then perhaps humanity can, for the first time, move beyond relying only on moral appeals such as:

“Everyone should cooperate.”

And begin to truly understand:

“What kinds of structures make cooperation a rational and sustainable outcome over the long term?”

I summarize the civilizational principle I currently understand in nine Chinese characters:

抓程序、做模式、求平衡

which can only be translated approximately as:

grasp the procedure, build the pattern, seek the balance.

This does not require everyone to accept the same religion,

nor does it require every country to adopt the same political system,

nor does it require the whole world to hold exactly the same values.

What it truly asks is:

When different people, different nations, and different civilizations possess different values and interests,

is there a more fundamental procedure of cooperation that allows differences to continue to coexist over the long term?

If there is not,

then the stronger AI becomes,

the more efficiently we may simply come into conflict with one another.

If there is,

then AI may, for the first time, be used to help humanity discover, simulate, test, and optimize these cooperative structures.

Mr. Kelly,

I have noticed that you have consistently maintained a rare form of long-term optimism about the future.

I do not wish to use catastrophic imagination to argue against that optimism.

In fact, what I most want to ask you is:

What civilizational structural conditions can truly support long-term optimism?

If technology naturally continues to expand the space of possibility,

then:

What determines whether humanity uses that ever-expanding possibility space for creation rather than destruction?

If AI gives humanity, perhaps for the first time, an almost indefinitely expandable capacity for cognition and action,

then:

Do we also need to upgrade the civilizational procedures that determine how these capabilities interact with one another?

Otherwise,

technological evolution may become increasingly advanced,

while the social relationships into which technology enters may still preserve the most ancient logic of the jungle.

Can these two continue to coexist indefinitely?

So what I truly hope to ask you is not for one answer, but for several questions:

First, when AI enters today’s human social structures, do you believe it will first amplify civilizational cooperation or competitive and harmful capacity? Why?

Second, if AI amplifies both for a considerable period of time, which force is likely to grow faster?

Third, if the capacity for mutual struggle and harm crosses a dangerous threshold first, is humanity more likely to be destroyed by an accident, or to be forced by survival pressure to upgrade its cooperative structures?

Fourth, could the evolution of technology itself eventually force a structural transformation in human social relations comparable to an evolutionary leap?

Fifth, if such a transformation is possible, should we wait for catastrophe to force it upon us, or can we begin in advance to treat the structure of human social cooperation itself as a serious object of scientific study and experimentation?

I do not know the answers to these questions.

Nor do I know whether my Procedural Determination Hypothesis is correct.

But I am increasingly convinced of this:

If humanity upgrades intelligence without upgrading the structures through which intelligences cooperate with one another, then “smarter” does not necessarily mean “more civilized.”

We may even encounter an extraordinary historical contrast:

Humanity may create the most intelligent tools in the history of civilization, while still using the oldest logic of the jungle to determine how those tools interact with one another.

That may be one of the things most worth worrying about in the age of artificial intelligence.

But it may also imply another possibility:

Artificial intelligence may ultimately do more than change what humanity is capable of doing.

It may force humanity to reconsider:

How should we act together?

If this latter path can be discovered proactively, instead of being found only after humanity is forced onto it by immense catastrophe,

then artificial intelligence may truly become a civilizational upgrade rather than merely a capability upgrade.

This is the question on which I most hope to hear your thoughts and criticism.

If the question itself is worth discussing, I also hope you might place it before more people who study technology, complex systems, civilizational evolution, and artificial intelligence, so that they too may challenge it.

If my direction is wrong,

please help point out where it is wrong.

If there is even a small part of it that deserves further study,

please help transform it from the reflections of an ordinary old man into a question that more people can examine together.

Thank you for reading this letter.

With sincere respect,

pen name: Jin Guyu
Former Senior Judge of a Local Court in China
Author of the Letters to Everyone in the World series

Email:——

（This is followed by the original Chinese text of the letter.）

致凱文·凱利：當人工智能闖入“叢林文明”，它增強的究竟是文明能力，還是人類互斗互害的能量？

尊敬的 Kevin Kelly 先生：

您好！

我筆名金穀雨，是一名來自中國的老人，曾任中國地方法院高級法官。

我關注您的思想，一個重要原因是：您並不只是把技術看作人類手中的某一件工具，而是把技術放進一個更長的演化過程之中，觀察它怎樣擴展可能性、改變人類，也改變文明本身。

因此，我特別想向您請教一個問題。

這個問題可以非常簡單地表達為：

當人工智能闖入今天仍然具有強烈“叢林邏輯”的人類文明以後，它首先增強的，究竟是人類創造文明的能力，還是人類互斗互害的能量？

如果答案主要是後者，那麼接下來會發生什麼？

人類會在某一次意外、誤判、競爭升級或者失控反饋中全部毀滅，還是會在巨大壓力之下走出另外一條文明演繹路徑？

這就是我寫這封信真正想請教您的問題。

這裡所說的“叢林文明”，並不是說今天的人類沒有文明。

恰恰相反。

人類已經創造了令人驚嘆的科學、法律、制度、藝術、倫理和技術。

但如果從社會合作結構的底層觀察，我們似乎仍然大量依靠一些非常古老的機制維持安全與秩序：

競爭；
威懾；
實力平衡；
利益交換；
控制；
防範；
相互猜疑；
以更強大的傷害能力阻止對方傷害自己。

國家如此。

企業如此。

政治集團如此。

很多人與人之間的關係也是如此。

於是就出現一個奇怪的文明悖論：

為了和平，我們不斷增強戰爭能力；
為了安全，我們不斷增強相互威懾能力；
為了不被控制，我們努力獲得控制別人的能力；
為了防止受到傷害，我們先獲得更強的傷害能力。

在過去，這種結構雖然危險，人類也許還有時間在一次次災難之後修正。

戰爭發生以後，人類建立新的規則。

危機發生以後，人類重新設計制度。

錯誤發生以後，人類慢慢學習。

但是AI可能改變這個時間尺度。

AI並不是進入一個空白的人類世界。

它進入的是我們已經存在的：

國家關係；

軍事體系；

商業競爭；

金融體系；

政治鬥爭；

信息傳播；

意識形態衝突；

個人慾望；

以及人與人之間已有的合作與對抗結構。

那麼真正的問題也許不是：

“AI本身是善還是惡？”

而是：

“當AI進入今天這種人類關繫結構以後，它會被什麼力量首先調用，並被什麼結構首先放大？”

如果AI使科研能力增長一百倍，

它當然可以幫助治療疾病。

但它也可以使武器研究能力增長一百倍。

如果AI使信息能力增長一千倍，

它當然可以幫助人類獲得知識。

但它也可以使欺騙、操縱、監視和認知戰爭的能力增長一千倍。

如果AI使決策和組織效率大幅提高，

它可以幫助人類合作，

也可以讓原本互相競爭的國家、企業和組織更加高效地競爭。

因此，我真正擔心的並不是：

AI突然變壞。

而是：

AI非常忠實、非常高效地幫助我們繼續按照今天的結構做事。

如果今天的底層結構本身存在問題，那麼一個越來越強大的AI，也許首先會使這個問題變得越來越強大。

這讓我想到一個非常樸素的問題：

技術進步的速度，與人類社會合作結構進步的速度，究竟是什麼關係？

假如技術能力呈指數級增長，

而人類處理利益、權力、安全、責任和相互信任的方式仍然只能通過戰爭、危機和災難慢慢糾正，

那麼兩者之間是否會出現一個越來越大的“文明速度差”？

技術每天都在變強。

但人類合作方式卻可能仍然停留在幾百年甚至幾千年來反覆出現的結構里。

如果這種速度差持續擴大，那麼AI帶給人類的首先可能不是一種全新的文明，

而是：

給舊文明安裝上一個越來越強大的發動機。

問題就變成了：

舊的方向，加上新的巨大動力，會把我們帶到哪裡？

我長期思考這個問題以後，形成了一個仍然需要檢驗的假說。

我把它叫作：

“程序定質假說”（Procedural Determination Hypothesis）。

它最簡單的表達是：

程序形成關係，關係形成結構，結構通過持續反饋形成動態平衡或失衡，並最終表現為一個系統的性質和結果。

也可以寫成：

程序 → 關係 → 結構 → 動態反饋 → 系統性質與結果。

我並不知道這個假說最終能否成立。

我更希望科學家尋找反例、檢驗它，甚至證明它錯誤。

但如果這個方向存在一部分真實性，那麼我們觀察AI時，也許不能只問：

模型有多聰明？

模型擁有什麼價值？

模型是否服從人類？

我們還必須問：

它進入了什麼程序？
與誰形成關係？
擁有怎樣的權限？
被什麼利益驅動？
受到怎樣的反饋？
最後進入了什麼樣的社會結構？

因為同一種技術進入不同關繫結構以後，完全可能產生不同的文明結果。

於是，我想把問題進一步交給您。

您長期思考技術的演化。

那麼從演化的角度看，您認為下面哪一種路徑更可能發生？

第一條路徑：AI增強文明能力

AI提高知識、創造、生產、醫療、教育和社會協調能力。

人類逐漸利用新的智能降低資源衝突，改善制度，提高透明度，擴展合作。

AI因此成為推動文明升級的力量。

這是我們最希望看到的未來。

第二條路徑：AI首先增強互斗互害的能力

國家使用AI對抗國家。

企業使用AI擊敗企業。

政治力量使用AI爭奪權力。

個人使用AI爭奪利益。

每一方從自己的局部角度看，都可能是理性的：

“別人正在增強，所以我不能不增強。”

結果卻形成一個越來越強的正反饋：

你增強——我增強；

我增強——你更加不安全；

你更加不安全——你繼續增強；

最終每個人都比以前更強大，

卻沒有任何人因此更加安全。

如果這種過程一直持續，

超級智能也許根本不需要“背叛人類”。

人類自己就可能藉助超級智能，把互斗互害能力提高到文明無法承受的程度。

如果第二條路徑真的出現，

那麼我最想請教您的，就是第三個問題：

人類隨後會怎樣演化？

一種可能是：

毀滅。

某個不起眼的錯誤；

某次軍事誤判；

某個AI agent未經預期的行為；

一次金融、能源、核武器、生物技術或者網絡系統的連鎖反應，

經過極高速度和巨大能力放大以後，

最終越過人類無法返回的臨界點。

那可能不是任何人的計劃。

甚至沒有真正意義上的“敵人”。

所有參與者都只是在自己的局部目標下做“合理”的事，

最終卻共同製造了毀滅性的整體結果。

但是，還有沒有另一種可能？

也就是：

巨大的技術壓力反而迫使人類社會合作結構本身發生一次根本演化。

人類第一次不得不認識到：

真正的問題不僅是：

誰擁有更強的技術；

誰掌握更多資源；

誰贏得競爭；

而是：

我們的關係究竟按照什麼程序組織？

也就是說，

技術演化最終是否會反過來迫使文明完成一次“社會合作結構的演化”？

如果過去：

資源不足推動合作；

戰爭推動國家形成；

貿易推動國際規則形成；

核武器推動人類建立某些新的戰略約束，

那麼人工智能是否可能成為一種更加強大的演化壓力：

迫使人類從以互害能力維持安全的文明，逐步走向以穩定合作結構維持安全的文明？

如果是這樣，

AI真正推動的最大革命，

也許最終不是機器智能革命，

而是：

人類社會合作方式的革命。

我因此提出另一個尚不成熟的概念：

“人類社會合作工程”（Human Social Cooperation Engineering）。

它並不是要把人變成機器。

相反，我只是想問：

既然人類已經能夠工程化地研究：

橋梁為什麼會倒塌；

飛機為什麼會失控；

網絡為什麼會崩潰；

AI系統為什麼會產生不期望行為，

那麼我們能不能也更加系統地研究：

社會合作為什麼會失衡？

能不能研究：

什麼樣的程序容易形成信任？

什麼樣的關係必然放大猜疑？

怎樣的權力結構會不斷自我強化？

怎樣的反饋會把所有參與者推向共同不希望看到的結果？

怎樣使每個參與者在維護自己合理利益的同時，不破壞整個系統繼續合作的條件？

如果這些問題能夠被觀察、比較、建模、實驗和證偽，

那麼人類也許第一次有可能不僅依靠道德呼籲：

“大家應該合作。”

而開始真正理解：

“怎樣的結構能夠使合作成為長期理性的結果。”

我把自己目前理解的文明做事原則壓縮為九個中國字：

抓程序、做模式、求平衡。

英文只能近似表達為：

grasp the procedure, build the pattern, seek the balance.

這不是要求所有人接受同一種宗教，

也不是要求所有國家接受同一種政治制度，

更不是要求全世界擁有完全相同的價值。

它真正追問的是：

當不同的人、不同國家、不同文明擁有不同價值和利益時，

有沒有一種更底層的合作程序，使不同者仍然能夠長期共同存在？

如果沒有，

AI能力越強，

可能只是讓彼此衝突得更有效率。

如果有，

AI也許第一次可以被用於幫助人類發現、模擬、檢驗和優化這些合作結構。

Kelly先生，

我注意到，您始終對未來保持一種罕見的長期樂觀。

我並不希望用災難想象反駁這種樂觀。

事實上，我更想請教：

真正能夠支撐長期樂觀的文明結構條件究竟是什麼？

如果技術天然不斷擴展“可能性”，

那麼：

什麼決定人類從這些越來越大的可能性空間中，選擇創造，而不是毀滅？

如果AI將使人類第一次擁有接近無限擴展的認知和行動能力，

那麼：

我們是否同時需要升級決定這些能力如何彼此發生關係的文明程序？

否則，

技術的演化可能越來越先進，

但技術所進入的社會關係卻仍然保持着叢林時代最古老的邏輯。

這兩者能夠長期共存嗎？

所以，我希望真正向您請教的，並不是一個答案，而是幾個問題：

第一，您認為AI進入今天的人類社會結構以後，首先會放大文明合作能力，還是競爭和互害能力？為什麼？

第二，如果AI在相當長時間內同時放大兩者，哪一種力量會增長得更快？

第三，如果互斗互害能力率先跨過危險臨界點，人類更可能因為一次意外而毀滅，還是會因為生存壓力而被迫升級合作結構？

第四，技術自身的演化是否可能最終倒逼人類社會關係發生一次類似“物種躍遷”的結構性變化？

第五，如果這種變化可能發生，我們是否應該等待災難迫使它發生，還是可以提前把“人類社會合作結構”本身變成一個嚴肅的科學研究和實驗對象？

我不知道這些問題的答案。

我也不知道我的“程序定質假說”是不是正確。

但我越來越確信：

如果人類只升級智能，卻不升級智能之間彼此合作的結構，那麼“更聰明”未必等於“更文明”。

甚至可能出現一個極大的歷史反差：

人類創造出了文明史上最聰明的工具，卻仍然用最古老的叢林邏輯決定這些工具彼此怎樣發生作用。

這可能是人工智能時代真正值得警惕的事情。

但它也可能意味着另一種可能：

人工智能最終不僅改變人類能夠做什麼，

還迫使人類重新理解：

我們究竟應該怎樣共同做事。

如果後一條路能夠被主動發現，而不是必須等到巨大災難以後才被迫走上去，

那麼人工智能也許真的能夠成為一次文明升級，而不僅僅是一次能力升級。

這就是我最希望聽到您思考和批評的問題。

如果這個問題本身值得討論，我也希望您能夠把它交給更多研究技術、複雜系統、文明演化和人工智能的人共同質疑。

如果我的方向錯了，

請幫助指出它錯在哪裡。

如果其中存在值得研究的一小部分，

也請幫助把它從一個普通老人的思考，變成一個可以由更多人共同檢驗的問題。

謝謝您閱讀這封信。

謹向您致以誠摯的敬意！

筆名：金穀雨
曾任中國地方法院高級法官
《致全球每一個人的信》系列作者

Email:——