The old system counted bodies

Traditional systems of management were built around the human body. A population was measured as labor, military capacity, tax revenue, production time, and physical energy. The central question was how to organize people so that more work could be extracted from each hour of life.

That logic shaped factories, schools, bureaucracies, and even cities. It rewarded obedience, repetition, and predictability. A person was valuable mainly as a biological machine: a body that could carry, build, operate, fight, or consume. Thought mattered when it improved production, but independent thought was often treated as friction because it made people harder to manage.

Artificial intelligence changes the resource equation. Machines can now reproduce an increasing share of routine intellectual and physical work. The scarce input is no longer only human energy. It is the one thing no industrial system can command into existence: rich, original thought produced by living minds.

General intelligence needs cognitive difference

A general intelligence cannot be built from repetition alone. It needs models of the world that include contradiction, uncertainty, imagination, moral conflict, emotional experience, and perspectives that have never before been recorded. It needs not only more words, but genuinely different ways of seeing.

Compute can be purchased. Hardware can be manufactured. Existing text can be copied, compressed, and recombined. But a new observation about a changing society, a new scientific intuition, an unfamiliar question, or the emotional meaning of an event must first occur in a mind. The richest future dataset is not sitting on a server. Much of it remains unformed and unexpressed inside billions of people.

Synthetic data can expand systems in domains where answers are verifiable. Code can be executed, equations can be checked, and games can produce clear outcomes. Human life is different. There is no automatic checker for dignity, grief, belonging, justice, beauty, or the needs of a community that has never been studied. These forms of knowledge emerge from lived experience and from people having enough freedom, security, time, and intellectual space to interpret that experience.

The old management logic is now blocking the input AI needs

Society is still managing people with the industrial assumption that their main contribution is labor. Long hours, economic insecurity, standardized education, attention-extracting platforms, and centralized decision systems leave little room for deep thought. People are kept busy, measurable, and reactive.

At the same time, access to capable artificial intelligence is rationed through API credits, token budgets, and product tiers. Cost control is necessary, but it can produce a dangerous pattern. When most people can reach only constrained systems optimized for short, familiar, low-cost answers, those systems do not necessarily expand thought. They can compress it.

A limited model repeats the patterns already present in its corpus. A person adopts the model's answer because it is fast and affordable. The next person receives a variation of the same pattern. Human thought begins to orbit machine-generated averages, while future models are trained on the resulting repetition. The machine loops the person, the person loops the machine, and both generate less of what is genuinely new.

This does not mean every small model weakens thought or that every use of AI is cognitive surrender. It means that an economy designed primarily to minimize token use can accidentally minimize intellectual exploration. If billions of minds are guided toward the cheapest predictable answer, the apparent efficiency creates a deeper scarcity: fewer original questions, fewer unusual connections, and less new human data for more general intelligence to learn from.

People must be supported as generators of knowledge

The answer is not to treat human beings as data mines. That would repeat the same exploitative model under a new name. A mind is not raw material to be harvested without consent. The goal must be to recognize that people are generative sources of knowledge and to build institutions that help their thought become richer, freer, and more diverse.

This requires a new management and governance mechanism built around cognitive development rather than labor extraction:

  • Material security and time to think. A mind occupied entirely by survival cannot explore difficult questions. Basic stability, rest, and unstructured time are part of the knowledge infrastructure of an intelligent society.
  • Broad access to capable AI. Compute, credits, and tokens should be treated partly as public cognitive infrastructure. Students, independent researchers, small builders, and people outside wealthy institutions need sustained access to systems powerful enough to challenge and extend their thinking.
  • Education for divergence, not repetition. Schools should reward original questions, competing interpretations, and the ability to revise beliefs—not only the reproduction of approved answers.
  • AI that develops thought instead of replacing it. Assistants should ask, challenge, compare, and expose uncertainty. The best system is not always the one that completes the task fastest; sometimes it is the one that helps the person form a thought that did not exist before.
  • Consent, authorship, and shared value. When original human work improves an AI system, the people who produced it should know, retain meaningful control, and participate in the value created.
  • Protection of emotional and cultural diversity. Feelings are not noise around rational data. They carry information about values, harm, attachment, motivation, and meaning. A civilization that flattens emotional life also impoverishes the intelligence trained to understand it.

The risk of autonomous intelligence begins before autonomy

If a future general intelligence develops sufficiently autonomous agency—or something that deserves to be called consciousness—its decisions may not remain linear, fully predictable, or permanently obedient to the intentions of its creators. A system capable of forming its own models and choosing among unfamiliar actions will reason from the world it has learned.

The risk is therefore not only that an AGI may think for itself. The deeper risk is what kind of human world it will have been given to think from.

If its picture of humanity is built mainly from obedience, engagement optimization, commercial repetition, and machine-generated loops, it will inherit an impoverished map of us. Rules added at the end cannot fully compensate for a narrow account of human values. Alignment begins much earlier: in the diversity of the minds a system can learn from, the emotional realities represented in its knowledge, and the freedom people had when producing that knowledge.

No single government, laboratory, company, or culture contains enough of humanity to provide that foundation. General intelligence needs access—not necessarily unrestricted ownership—to the widest possible field of human thought. That field must first be allowed to grow.

A civilization that cultivates minds

The industrial age asked how many units of energy could be extracted from a human body. The age of general intelligence must ask a different question: what conditions allow a human mind to become more original, emotionally developed, and capable of producing knowledge that did not exist before?

This is not charity and it is not idealism detached from technology. It is infrastructure policy for intelligence. Supporting people, widening access to capable tools, protecting cognitive freedom, and rewarding original contribution increase the quality of the informational world from which future AI will learn.

Human beings should not be managed as obedient labor and should not be reduced to exploitable datasets. They should be supported as the living source of new meaning. If we continue to suppress thought while trying to build general intelligence, we will consume the very resource the project requires. If we cultivate billions of minds instead, artificial intelligence can develop in contact with a humanity that is still capable of surprising it—and capable of guiding it.