Scarcity: What Changes and What Remains Constant Before and After the Singularity

Table of Contents

In early August 2026, leading figures in the AI field, including Sam Altman and Elon Musk, successively announced: we are in the midst of a “Singularity moment.”

It’s not approaching, not imminent—we are already in it.

The most direct evidence is that in just the past month, several long-standing, unresolved problems in mathematics have been solved consecutively using mainstream AI models.

For instance, on August 1st, OpenAI announced that using an internal model codenamed Astra, they delivered ten achievements in mathematics and theoretical computer science in one go. These included three Erdős problems—problems from the collection of unsolved problems left by the legendary Hungarian mathematician Paul Erdős. Some of these challenges were completely solved, some saw decisive progress, and for others, AI constructed counterexamples, disproving well-known conjectures [1].

Any one of these problems, if solved by a human mathematician, would be enough to earn prestige, even major awards. Yet, the total computational power spent by AI to solve these 10 problems amounted to only $2,000.

An “earthquake” is insufficient to describe AI’s impact on mathematics—this is a “phase transition”: the methods of mathematical research will be fundamentally different from now on.

This is only the field of mathematics. Other theoretical and engineering research, as well as experiments, biology, and medical research directly involving AI, are also advancing rapidly. It is truly exciting to anticipate the astonishing results that will be achieved once new-generation models are fully applied to front-line scientific research.

The reality is, even if AI technology stopped progressing today, simply widespread adoption of existing capabilities would be enough to bring about dramatic changes in society.

This article will look ahead to the world after the Singularity.

In 2012, Jeff Bezos offered a profound insight: “People often ask me, ‘What’s going to change in the next 10 years?’ And I almost never get the question, ‘What’s not going to change in the next 10 years?’ The second question is actually the more important one, because you can build a business strategy around the things that are stable over time” [2].

So, what will remain unchanged even after the Singularity?

Specifically, what will continue to be scarce, or even become scarcer?

What Will Change Dramatically After the Singularity? #

People looking back at our era a hundred years from now will find many things incredible, much like we view bloodletting today—

That we allowed humans to drive. That we permitted an animal prone to fatigue, road rage, and distracted driving (using a phone) to operate a two-ton iron box at speeds exceeding a hundred kilometers per hour, treating over a million human lives annually as a normal societal cost [3]?
That our healthcare was “fix it when it breaks.” No monitoring in daily life, only visiting the doctor when symptoms appear, and everyone taking the same medicine—a medicine whose efficacy was validated on thousands of “average people,” but you are not “average.”
That we considered aging a normal phenomenon. People a hundred years from now will likely treat aging as a treatable disease that can be intervened with. They will view our resignation to aging much like we view ancient people’s resignation to smallpox.
That merely for taste, we subjected tens of billions of sentient animals to a lifetime of suffering in factory farms, only to be slaughtered.
That we made all children sit in the same classroom, learning the same content at the same pace, and then used a single multi-hour exam to determine a person’s entire life’s destiny.
That many, even most, people in our society had to toil and expend their lives simply to earn the right to exist…

They will say, “Humans shouldn’t have lived such lives.”

And it is our responsibility to accelerate the change of such lives. But for now, let’s consider: what will not change?

Conflict and Politics: The Eternal Game #

First, conflicts between people will not change, nor will the essence of politics.

Some have fantasized that since AI is smarter than everyone, without personal biases or susceptibility to bribes, why not let AI govern nations? Couldn’t we just input all relevant information to AI and let it make the best arrangements?

This is not merely technically infeasible for now; it’s mathematically impossible.

We previously mentioned Arrow’s Impossibility Theorem, proven in 1950 by the American economist Kenneth Arrow, who later won the Nobel Memorial Prize in Economic Sciences [4]. This theorem states that given at least three alternatives and at least two individuals, no aggregation method can combine individual preference orderings into a single “social preference ordering” while satisfying several minimal reasonable conditions—for instance, that no single individual can dictate the outcome.

In short, society will always have differing priorities, orderings, and needs between individuals—there is no mathematical method that can reconcile our conflicts in a universally satisfactory way.

As long as good things are finite and can only be distributed to some people first, there will inevitably be a need to “negotiate” who comes first and who comes next. Perhaps it’s negotiation, compromise, or voting and alliance-building. Use words rather than force, if possible, but the precondition is that you possess the strength to use force—this is politics. AI will not change this situation; mathematics does not allow it to change.

Conflict will not disappear, strategic interaction will not disappear, nor will uncertainty; and on many issues, human experience and judgment remain the ultimate arbiters.

Whether a movie is good, a dish is tasty, or a product sells well—these questions lack formal criteria. Perhaps some people say yes, while others insist no; perhaps you once thought it was good, but now you don’t. The judgment of “good or bad” can ultimately only be made by humans.

In other words, even if an individual doesn’t act as a producer or creator, their role as a mere consumer or critic is irreplaceable.

Unchanging Scarcity: The Value of Positional Goods #

But perhaps you have a grander ambition; in that case, you will be concerned with another constant: “scarcity.”

Some say that when AI significantly boosts productivity and material abundance, we will enter a “post-scarcity society”… However, there is a type of scarcity unrelated to material abundance; it is defined by “position.”

In 1976, British economist Fred Hirsch introduced the concept of “positional goods” in his book Social Limits to Growth [5]: the value of some things comes not from what they are, but from where they rank—and “rank,” by definition, is finite. We call this situation “positional scarcity” or “ranking scarcity.”

No matter how advanced sports technology becomes, there will always be only one gold medal for an Olympic event. The World Cup occurs every four years, and the final is broadcast live worldwide, yet there are only tens of thousands of seats at the venue—the experience of “being there” is expensive, and it will only get more expensive. There are only so many houses in the best locations in “first-tier” cities, only a few seats in first class on a plane, and only… 10 universities in the national Top 10.

Prestige is also positional, limited by the scarcity of attention. You only truly follow a few celebrities. If your channel ranks 12,534th on the Douyin famous channels leaderboard, yours is not a “famous” channel.

We’ve discussed “Baumol’s cost disease”: technological progress makes things that can be made more efficient increasingly cheaper, while things that cannot be made more efficient become relatively more expensive. Thus, the Singularity will only exacerbate positional scarcity.

In extreme cases, if AI makes everything that can become cheaper free, then the remaining things that cannot become cheaper will carry almost all of the scarcity premium.

As long as technology is the answer to a problem, the thing corresponding to that problem will become widespread; widespread adoption means devaluation.

Whether it’s a brand or prestige, to avoid devaluation, you must occupy a unique position within the finite total attention of all humanity.

Scarcity Enduring the Singularity: Three Core Values #

When I was little, I read “The Ballad of Mulan,” and the lines “Last night I saw the conscription notice; The Khan is raising a great army, In twelve scrolls of military records, Each scroll carries father’s name.” I didn’t feel any sympathy for Mulan’s elderly father; instead, I thought he was formidable: specifically chosen to fight, and his name on every scroll—what immense prestige!

Of course, the true meaning of “Each scroll carries father’s name” is forced conscription. What you desire is not to be conscripted, but for others to need your participation in important matters, and for you to have the right to decline.

In fact, everyone holds such value. For example, when relatives or friends host celebrations or funerals, your presence is an endorsement. You are an irreplaceable, intrinsically valuable, unique individual: your time is finite, thus your presence is scarce. The Singularity will only intensify this.

But “one-of-a-kind” might not be enough for you—after all, every snowflake is also one-of-a-kind—perhaps what you truly desire is a greater scarcity… ideally, to occupy a position that many wish to hold but cannot.

There are three such positions, all capable of withstanding the sweep of the Singularity: the blank space, the guarantor, and the connector.

Core Value One: Filling the Blank Space #

The first position is filling the blank space. This “blank space” refers to what we’ve discussed as “residual control rights.”

Contracts can only stipulate so much. A company is a machine seeking opportunities to make money, not an organization that strictly adheres to contractual instructions. AI is good at answering questions, but it doesn’t have the authority to set problems for humans.

Where AI cannot decide, in the blank spaces of agreement clauses, beyond the precipice of all hard constraints—that is where you apply your talent and set problems. You must weigh pros and cons yourself, make choices based on your own judgment, and influence others through your initiative.

Should this project be done? To what extent is it considered successful? Which of two reasonable proposals should be chosen? Whose responsibility is a new situation not covered by the rules? The work in these blank spaces is not about “getting it right,” but about “making a judgment.” You make a decision based on your “feel” and then bear the consequences of that decision.

Imagine a highly automated shipyard, where blueprints are system-generated and welds are machine-made. The hull and propulsion each have a set of specifications, and both sets stop at the boundary: the hull side says, “beyond this line belongs to propulsion”; the propulsion side says, “beyond this line belongs to the hull.” Both sets of specifications are correct, but combined, there’s an inch-long gap in between.

Ask the system what to do about this inch, and it will reply: “If you want to be a bit safer, do it this way; if you want to save a bit of money, do it that way; or you can handle it according to industry practice.” But this is a new model; there is no industry practice. Who knows if that little bit of extra safety is worth that little bit of extra money?

Finally, an old engineer made the call: this inch belongs to the hull, add two reinforcing ribs, costing an extra one hundred and ten thousand.

“What if it’s wrong?” someone asked. The old engineer stamped his seal: Process Responsible Person, Chen [X].

Ultimately, we can only trust someone’s subjective judgment—because only humans can take accountability.

Core Value Two: Being a Guarantor #

This leads to the second position: being a guarantor.

If you merely possess wealth, you are just a consumer. But if you put what you own up as collateral, you are “doing something.”

AI prepares all financial statements for the accountant, but ultimately the accountant must sign off, putting their license, career, and personal wealth on the line. You must “have something” for your collateral to be valuable.

At the height of the 2008 financial crisis, Warren Buffett invested $5 billion in Goldman Sachs—for many in the market, this money was a reassurance: “Buffett is in.” That name was immensely significant.

Taking the Astra math achievements as another example, those results were produced by AI, and in principle anyone could achieve them with AI—but when OpenAI dared to compile them into a paper and publish it under the company’s name, you are mortgaging your reputation. If you dare to publish, the AI community will dare to believe, and the mathematical community will dare to spend time verifying. If they discover AI made a mistake, your reputation will be severely damaged.

When we talked about “self-constraint,” we said: the end of the chain of responsibility must be the party that can be harmed.

Being able to be harmed is a privilege humans have over AI; being worthy of being harmed, however, is a qualification not everyone possesses—you must have a reputation to ruin, assets to lose.

An ordinary person can vouch for a friend; Buffett can endorse Goldman Sachs. Carefully accumulate your reputation and assets; do not spend them lightly.

Core Value Three: Being a Connector #

The third position is being a connector.

Everyone exists within social relationships: you are the child of some, the parent of others, the spouse of one, and a close friend to several. To the world, you might be replaceable, but to them, you are not. This is why even an elderly person who is incapacitated and can do nothing but lie in bed still has value—their value lies not in what they can do, but in “who they are to whom.”

By the same logic, no matter how thoughtful or comprehensive AI companionship and care may be, it cannot take away your value as a friend or family member—because the very definition of that value includes “being you.”

The “structural hole” we discussed is also a form of relationship value.

Some groups have no connection between them, leaving a hole in between. Whoever bridges this hole—known by both sides, neither side knowing the other—occupies a naturally valuable position: information passes through them, opportunities pass through them, and they can bring the two sides together. So why isn’t AI, which excels at processing information, filling these structural holes?

Here’s a key concept called “incommensurability.” In short, it means these two things have no common denominator and cannot be measured by a common standard.

For example, fame and money, security and freedom, career and family, a child’s happiness and a child’s future—these things can be traded off, and we do so every day, but there is no “exchange rate” between them. How much fame is equal to ten thousand dollars? That number doesn’t exist.

The British philosopher Isaiah Berlin’s term for this is “value pluralism”: the world of values is full of such relationships [6]. In essence, value incommensurability and Arrow’s Impossibility Theorem are two sides of the same coin: the fundamental reason why Arrow-like difficulties are inescapable is that there’s no common conversion unit between the priorities of different people—whose grievance is more important, Zhang San’s or Li Si’s satisfaction? No one can set an official exchange rate.

Since two values are incommensurable, there is no answer that can be mechanically calculated using a common ruler and accepted by all parties at first glance. Yet, we still have to make decisions. In 1997, American philosopher Ruth Chang specifically argued that incommensurability does not mean undecidability [7].

So who makes the decision? If you choose a solution for two parties, why would they accept it?

In fact, traditional Chinese society long had its own solution for structural holes, known as the zhongren (middleman/intermediary). Contracts during the Ming and Qing dynasties—for buying land, selling houses, borrowing money, becoming an apprentice—almost invariably listed zhongren. Researchers found a land sale contract from the eighth year of Tongzhi (1869) that listed as many as twenty-four zhongren [8]. The role of these zhongren was to mediate, witness, guarantee, and reconcile.

The Great Qing had laws and yamen (government offices), but the land seller’s internal ledger recorded ancestral property, face, and village gossip—“my ancestral land was sold under my watch,” a pain without a price; the land buyer’s ledger recorded money, land fertility, and harvest. These two sets of accounts are incommensurable; even if you invited AI, it would be useless.

The zhongren could make the transaction happen because they were familiar with both parties, trusted by both parties, and, more importantly, could impose consequences on both parties—the seller knew that if they were unfair, they would lose face in the village from then on; the buyer knew that if they messed up, no one in the market would ever ask them to be a zhongren again.

He would say, “I think this amount is fair. If there are any issues later, you can come to me.” Both sides accepted this statement.

They didn’t accept the number. They accepted him: the fact that he simultaneously put his own reputation in both circles on the line.

Finally, let’s imagine. If you traveled a hundred years into the future, or lived that long, you would be surprised to find that these things are still around.

Marital relationships still exist, and people are still arguing. They’re still arguing about the same things: whose mother is more important, yours or mine; how your career compares to my contributions. And they still can’t resolve it—because these accounts are incommensurable.
Dinner parties still have seating arrangements. There’s still only one main seat, and there are still rules about who takes the first bite.
Wedding guests still need to attend in person. The body can only be in one place at a time, which is why that celebratory drink is so precious.
Signing major contracts still requires finding someone trusted by both parties to mediate, perhaps reverting to the traditional name zhongren.
Parliaments are still arguing, and they’re still arguing about the same things: who comes first, who gets more, based on what grounds…

In this scenario, even if everyone receives universal basic income, people’s social lives probably won’t undergo many qualitative changes. Think about it: aren’t the retired seniors in the system today already living a “post-Singularity” life?

The Singularity takes away the premium on ability; the remaining value is entirely in position.

In twelve scrolls of military records, Each scroll carries father’s name—why do I love this line so much…? In short, you don’t want to be an invisible person; you want to be a prominent figure in society, a character, a local hero.

【Epilogue Poem】

Ask what you deem precious, Ask what you hold dear. A hundred crafts can be replaced, But one’s promise binds the self. Beyond the contract, a space remains; Among the seats, one is empty. In the world’s myriad scrolls, Each awaits your name.

Citations

[1] OpenAI. “Ten Advances in Mathematics and Theoretical Computer Science.” August 1, 2026. https://openai.com/index/ten-advances-in-mathematics/

[2] Amazon Web Services. “2012 re:Invent Day 2: Fireside Chat with Jeff Bezos & Werner Vogels.” YouTube video, November 29, 2012.

[3] World Health Organization. Global Status Report on Road Safety 2023. Geneva: WHO, 2023.

[4] Arrow, Kenneth J. “A Difficulty in the Concept of Social Welfare.” Journal of Political Economy 58, no. 4 (1950): 328–346.

[5] Hirsch, Fred. Social Limits to Growth. Cambridge, MA: Harvard University Press, 1976.

[6] Berlin, Isaiah. “The Pursuit of the Ideal.” In The Crooked Timber of Humanity: Chapters in the History of Ideas, edited by Henry Hardy. London: John Murray, 1990.

[7] Chang, Ruth, ed. Incommensurability, Incomparability, and Practical Reason. Cambridge, MA: Harvard University Press, 1997.

[8] Guo, Ruijun. “Who Is the Intermediary: The Choice of ‘Zhongren’ in Huizhou Local Society in the Qing Dynasty.” Chinese Social Sciences Today, June 25, 2025.