Appendix I · Two Stages, Not One Future

Appendix I: AGI and ASI—Two Eras, Not One Future

The most common—and most costly—mistake in thinking about the future is to fold “the AGI era” and “the ASI era” into a single picture: telling people who have just lost their jobs that “everything will be free someday,” while waving off deeper change with “AI just writes code.” Between the two lies no smooth upward slope but a fault line. AGI first changes the price of cognitive labor, while bills, property and institutions stay in place; only when the rules of production, deployment and distribution are rewritten together could ASI erase the constraints of daily life itself.

The distinction in one line: AGI makes “doing” cheap, but it does not make “owning” cheap. As the marginal cost of cognitive services collapses, whatever stays scarce—land, energy, compute, distribution channels, credentials and human trust—absorbs most of the gains. That is where the pain of the transition comes from. The ASI branch is about changing exactly these bottlenecks of ownership.

STAGE A · Transition and Breach

New Wine in Old Wineskins

Window to watch: late 2020s to early 2030s (author’s estimate, for digital cognitive ability at skilled-human level)

Capital, firms, nations, money and debt keep running, but the production function inside them has been rewritten: cognitive work that once took ten people may now take one person and a set of agents. Output grows, yet entry-level jobs and labor income shrink first; institutions still tax, pay pensions and sort talent on the old assumption that “people live on wages.”

How it feels: things keep getting smarter, and the bills don’t shrink by a cent; opportunities seem to multiply, yet the bar keeps rising.

STAGE B · Phase Change and Rebuilding (Scenario Branch)

The Vessel Itself Is Recast

Rapid-improvement scenario: the capability could arrive in the 2030s; when it reaches daily life depends on deployment and distribution—later, and uncertain

This appendix adopts one explicit branch: advanced production technology matures, autonomous engineering is deployed at scale, and basic living needs are universally guaranteed. Only under this joint premise could food, shelter and care shift from “goods you must buy” to “infrastructure you receive on demand,” and only then would wage labor leave the center of how survival is distributed.

How it feels: survival anxiety recedes, and competition over meaning, status and attention rises to the surface.

Mechanisms Compared: Where the Two Stages Diverge

Dimension	AGI Transition	ASI Post-Scarcity Branch
Essential nature	An extraordinarily powerful tool: within existing physical and social rules, it performs cognitive tasks at the level of skilled humans or even experts.	A rebuilder of reality: it does not rewrite the laws of physics, but it changes how matter, energy and engineering are put to work—and thus the cost of getting anything done.
The real bottleneck	Power and grid connections, chip capacity, robots and physical manufacturing, permits and liability.	Energy at scale, materials cycles, the safety limits of autonomous engineering—and who has the right to distribute output.
Price signals	Cognitive services plunge in price; premiums rise on scarce physical goods and locations; gains concentrate among those who hold the bottlenecks.	The price of basic necessities approaches the cost floor; scarcity shifts to location, attention, human time and historical originals.
Social machinery	Firms, states, taxes and social insurance come under severe strain and are patched frantically, yet they remain the backbone of distribution.	Distribution rules must be rewritten. This work treats unified algorithmic governance (a singleton) as a setting of this branch, not as an inevitable result of rising intelligence.
Human condition	People lose their bargaining leverage as workers; fear of replacement, debt risk and identity crisis pile up, and survival anxiety peaks.	Survival risk falls, and autonomy, meaning and psychological resilience become the central questions: this may be ontological freedom—or nihilism.
Most dangerous misjudgment	Assuming that “greater efficiency” automatically means “better lives.”	Assuming that “technically possible” means “available to everyone.”

Three Clocks Out of Sync

The pain does not come from technology moving too fast; it comes from three clocks running at different speeds. Capability advances by the month, deployment by the year, and institutions by the decade or the generation. The gap between the clocks is time a society must pay for in unemployment, debt and division.

CLOCK 1 · Months

The Capability Clock

Leaps in model performance on benchmarks and real tasks. It determines what can be automated in principle—not who loses a job tomorrow.

CLOCK 2 · Years

The Deployment Clock

Inference costs, data centers and grids, robot hardware, industry integration, accident liability and certification. It determines when capability becomes a service that can be used at scale.

CLOCK 3 · Decades

The Institutional Clock

Tax bases, social insurance and pensions, degrees and professional licenses, labor law and property. It determines how new output is shared—and whether ordinary people experience prosperity or abandonment.

The AGI transition is an era in which the capability clock runs far ahead of the institutional clock. For the ASI branch to deliver post-scarcity, autonomous engineering must compress the deployment clock—but no technology can speed up the institutional clock on its own. Only political choices can move it.

Conditions for the Leap: What Must Be Crossed to Get from AGI to the ASI Branch

Condition 1: Research automation closes the loop

AI not only proposes hypotheses but designs and runs experiments, reads the results and iterates. The bottleneck moves from “ideas” to “experimental throughput”: lab robotics, sequencing and synthesis capacity, clinical and safety validation.

Condition 2: Physical production can expand itself

Robots can build, maintain and deploy robots and factories at acceptable cost. As long as physical expansion depends on scarce skilled workers and long construction cycles, “supply on demand” in the endgame stays a demo.

Condition 3: Energy and materials constraints loosen dramatically

Whether through controlled fusion, cheap storage or another route, what matters is not the name of the breakthrough but that the all-in cost of energy and key materials keeps falling—and that access is fair.

Condition 4: Distribution rules are rewritten

Even if production costs collapse, if output and compute are owned by a few, humanity may end up with “abundant technology and scarce lives.” Universal provision requires new forms of ownership, taxation or allocation.

Condition 5: Safety and governance are verifiable

Before autonomous systems run the physical world at scale, society needs mechanisms that can audit them, correct them and hold them accountable. Without this, the more the first four conditions succeed, the greater the risk.

Miss any one of the five conditions and you won’t get the endgame this appendix describes, but a hybrid state instead: pockets of abundance alongside widespread anxiety.

Misreading 01

Misreading 1: Explaining the Transition with the Endgame

Telling someone who has just been laid off, with a mortgage still to pay, that “everything will be free in the future” is neither accurate nor responsible. The real constraints of the transition are cash flow, debt maturities and the window for retraining; an endgame vision is no substitute for designing protections for this generation.

Misreading 02

Misreading 2: Dismissing the Endgame with the Transition

Seeing that today’s AI still makes mistakes and robots are still clumsy, and concluding that deep change won’t happen, is just as dangerous. Capability and deployment often move slowly and then quickly; what to watch is whether the bottlenecks are loosening, not whether a single demo is flawless.

“The AGI era is an elimination round: the old order scrambles to patch its holes, and people are stripped of their value as tools in the afterglow of the capital machine. If the ASI branch comes true, it is more like a change of question: the old contradictions are not solved by humans—the cost coordinates they rested on are rewritten.”

—— Editor’s note to the appendix

Note: this appendix is a scenario analysis. Time windows follow the author’s estimates on P02; the ASI picture of daily life applies only if the joint premises above hold, and it is neither a forecast nor advice.