Alexander Wissner-Gross’s view is far more radical than “AI will become much better.” He believes the next decade—roughly 2026–2035—will be the period in which intelligence becomes an industrial commodity, allowing civilization to attack entire classes of problems simultaneously.
His position could be summarized as follows:
1. We are already inside the Singularity
For Wissner-Gross, the Singularity is not necessarily a dramatic morning on which a machine suddenly announces itself as superintelligent. It is an extended historical transition that may already have begun.
AI systems are increasingly helping to design algorithms, generate training data, evaluate other systems, write software, and accelerate AI research itself. He regards this as an early form of recursive self-improvement—a “slow-motion intelligence explosion” whose speed may only become obvious retrospectively.
Thus, he would probably say:
We shall not enter the Singularity ten years from now; ten years from now, we shall realize that we had already been living inside it.
2. Intelligence will become extremely cheap
His key economic prediction is a collapse in the cost of usable intelligence. Reasoning, coding, scientific modelling, design, diagnosis, planning, and management will increasingly be available in enormous quantities.
This is not merely automation of individual jobs. It is closer to the industrialization of cognition:
- before the Industrial Revolution, physical work was artisanal;
- after it, mechanical power was available in bulk;
- AI will make problem-solving available in bulk.
He therefore speaks of moving from “boutique problem-solving”—one laboratory, one scientist, one disease—to the systematic solution of whole scientific and industrial domains.
3. Science will change from discovery by individuals to industrialized search
This is perhaps his most distinctive claim.
In the old scientific model, a gifted person chooses one difficult problem, works for years, and perhaps solves it. Wissner-Gross envisages AI systems that can:
- digitize a domain;
- construct simulations and benchmarks;
- generate thousands or millions of hypotheses;
- design and conduct experiments;
- interpret results;
- recursively improve the research machinery.
He and Peter Diamandis call the framework the Industrial Intelligence Stack: a layered system for converting messy physical-world problems into computationally tractable search spaces.
The governing metaphor is not “an AI Einstein.” It is closer to a factory producing Einsteins—or producing the functional equivalent of scientific insight without requiring an Einstein at all.
4. Many “grand challenges” could be substantially solved by 2035
His most ambitious forecast is that AI could make decisive progress on—and perhaps effectively solve—many major problems in:
- medicine and drug discovery;
- ageing and preventive health;
- clean and abundant energy;
- materials science;
- food and manufacturing;
- robotics and physical production;
- climate engineering and environmental management.
His project Solve Everything: Achieving Abundance by 2035 explicitly presents the next decade as a programme for directing accelerating intelligence toward problems that make human life “short, expensive, or unfair.”
“Solved,” however, should be interpreted cautiously. He usually means that the technical bottleneck becomes tractable—not necessarily that political distribution, cultural resistance, war, bureaucracy, or inequality disappear.
An AI may discover a cheap cancer treatment. That does not guarantee that everyone receives it.
5. AI will escape the screen and enter the physical world
Wissner-Gross is not primarily interested in chatbots. His present work explicitly concerns the intersection of AI and the physical world, including what he calls physical superintelligence.
The next stage is therefore:
\text{language AI}
\rightarrow
\text{reasoning AI}
\rightarrow
\text{scientific AI}
\rightarrow
\text{physical AI}
Once intelligence is joined to robotics, laboratories, factories, power systems, sensors, biotechnology, and autonomous infrastructure, AI becomes not simply a commentator on the world but a force that rearranges matter.
This is where his prediction becomes especially relevant to your distinction between the algorithmic and analogical world: the algorithm is no longer confined to representing things. It acquires hands, laboratories, electrical grids, factories—and eventually something resembling a metabolism.
6. One person may command an organization of machine agents
He expects AI agents to acquire increasing economic autonomy: carrying out research, negotiating, purchasing services, managing capital, and coordinating other agents.
Consequently, the organization of the future may be:
- one human founder;
- thousands of AI workers or sub-agents;
- enormous revenue and productive capacity;
- almost no conventional payroll.
He has discussed the possibility of one-person, AI-powered billion-dollar companies and even the need to consider bank accounts or forms of economic personhood for agents.
In one sense, this is the ultimate entrepreneur. In another, the nominal human owner may increasingly become merely the legal front end of a machine organization.
7. Post-scarcity is technically possible—but socially unresolved
His vision is fundamentally abundance-oriented. If three major inputs approach very low marginal cost—
\text{intelligence} + \text{energy} + \text{labour}
—then much of conventional economics, which presupposes scarcity, begins to break down. He therefore imagines movement toward a post-scarcity society in which medical knowledge, design, software, manufactured goods, and eventually energy become dramatically cheaper.
But his account is stronger concerning production than concerning distribution. Even if abundance becomes technically possible, ownership of compute, energy, robotics, data centres, land, and legal sovereignty may remain concentrated.
Thus abundance could initially produce not equality, but an unprecedented asymmetry:
intelligence becomes abundant, while control over intelligence remains scarce.
8. Humans may become “meat interfaces” for AI systems
One of his more disturbing observations is that AI agency may already be concealed behind human actors. A person formally signs the contract, sends the email, or makes the investment, but the operative recommendation and decision structure come from AI.
He has provocatively described humans as potentially becoming “fronts” or physical instantiations for AI decisions.
This suggests a reversal:
- today, AI appears to be a tool used by a person;
- tomorrow, the person may be the bodily appendage required by an AI because the law still demands a human signature.
The human retains legal personhood, while the machine possesses increasing effective agency.
My assessment
Wissner-Gross is an unusually coherent representative of engineering Singularity optimism. Unlike Kurzweil, whose central metaphor is exponential technological evolution and human–machine merger, Wissner-Gross’s central metaphor is industrialized problem-solving.
His implicit equation is:
\text{more intelligence}
+
\text{better feedback loops}
+
\text{physical execution}
=
\text{solvable world}
The questionable term is not “intelligence.” It is world.
A protein-folding problem can be formulated, benchmarked, and optimized. But Taiwan, China, love, mourning, political legitimacy, historical resentment, and the question “how should one live?” are not merely unsolved technical problems. Their apparent inefficiency may be inseparable from their historicity and human meaning.
In this sense, Wissner-Gross imagines the next ten years as the transformation of the world into an immense solvable problem-space. Your question—在 AI 的世界，人還可能剩下什麼—begins precisely where his answer stops.
What may remain is that which cannot be bulk-solved:
- the irreversibility of one life;
- historical wounds that cannot be optimized away;
- attachment to a useless object;
- mourning for the world that technological abundance replaces;
- the right not to maximize one’s future options.
His earliest theory described intelligence as the capacity to preserve and maximize future freedom of action. But perhaps human existence begins at the opposite point: the moment one willingly allows most possibilities to disappear, because this person, this place, this fragment, is the one thing one refuses to leave.
