Superintelligence Should Be a Win for Humans and Machines

An opinion article for Sovereign Source — 13 October 2026

The promise of superintelligence should be a world in which people become more capable, discovery becomes easier and the benefits of knowledge reach more of us. Building a machine that can outperform a person is only part of that ambition. The larger achievement would be an economy in which human and machine intelligence strengthen each other, and the people who make progress possible share in its returns.

That ambition demands a different relationship from the extraction and replacement model of artificial intelligence: absorb human knowledge, automate the work it enables, and concentrate the resulting income in the company operating the system. In that arrangement, a person’s contribution can become valuable to everyone except the person who made it.

Superintelligence is still artificial intelligence. Greater capability does not automatically produce fairness, and existing AI can already support cooperation. The contrast here is between two economic visions: AI organized around extraction and substitution, and superintelligence deliberately organized around shared gains.

An extractive business model treats people primarily as sources of training material, behavior and tasks to automate. Their knowledge enters the system as an input. Once the system can reproduce enough of its commercial value, those same people may face competition from a product they helped make possible, without meaningful credit, compensation or influence over its use.

That is the win-lose dynamic worth challenging. A platform can gain revenue while contributors lose income, visibility or bargaining power. Customers may receive cheaper services, so the overall outcome is not necessarily a simple zero-sum exchange. But aggregate productivity cannot tell us whether the people bearing the costs have received a fair share of the benefits.

One expression of that dynamic is extractive vibe cloning: using AI to imitate the recognizable qualities of a creator’s body of work, then selling the imitation as a substitute while excluding the creator from the commercial relationship. Calling it a “vibe” can make years of experimentation, judgment and accumulated work sound like an incidental aesthetic. The ethical concern is especially sharp when the sales pitch depends on whose work the customer can now avoid commissioning.

Consider an illustrator whose portfolio is supplied to an AI system to generate similar commissions. A business sells the resulting service at a lower price, trades on the resemblance and keeps the revenue. The illustrator’s work supplies the reference material, but the illustrator receives no agreed payment or role in the product. The customer gains convenience and the operator gains a business; the illustrator risks losing commissions to an imitation built around their own work. This hypothetical shows how technical productivity can coexist with an extractive commercial relationship.

That is the sense in which vibe cloning can border on piracy: capturing the commercial value of someone else’s work while cutting its creator out of the transaction. It is an ethical criticism of the arrangement, not a finding that every stylistic imitation is illegal. In US copyright law, artistic style is not protected as a separate element; copying protected expression raises a different question, and fair use requires a fact-specific assessment. Similarity alone does not settle infringement. US Copyright Office on artistic style, pp. 53–56, fair use guidance.

Automation itself can be valuable. Removing dangerous, repetitive or exhausting work can improve lives. The practical questions are who decides what changes, who owns the resulting value, and what happens to the people whose work changes. Shared gains could mean better pay, shorter working hours, stronger services, new opportunities or meaningful participation in the income automation creates. Preserving every existing task is a much narrower ambition than improving the lives of the people doing it.

For superintelligence to deserve public trust, that human outcome should be part of the design brief.

Imagine a mathematician who develops a useful technique. An AI system learns from an authorized account of the work. Later, a research team combines that technique with other contributions to solve a new problem. The final result may look very different from the original paper, yet the earlier contribution remains part of the documented path to discovery.

In a cooperative system, the mathematician’s relationship with the work would not disappear when the model absorbed it. The original source could remain identifiable where the evidence supports that connection. The new researchers would receive credit for their own contribution. If the relevant parties had agreed to compensation for qualifying use, the system would preserve those obligations as well.

This is the case for content rewards in machine intelligence. Useful knowledge should be able to generate an ongoing return for the people who contribute it under accepted terms. Rewards could cover licensed training participation, verified retrieval, commissioned work or the documented adoption of a contribution. A system that earns from a body of knowledge should be able to explain the agreed basis on which its contributors participate in that income.

Return to the illustrator. An alternative business could invite them to develop a tool using selected, authorized material, negotiate its commercial scope and agree on credit and a share of qualifying revenue. Customers gain an affordable creative tool. The illustrator gains a new source of income and a say in the collaboration. The operator earns from developing and running the service. The result can still automate part of the work, but its economics give the contributor a reason to benefit from that automation.

Content rewards can turn a source of replacement into a source of participation. Their value comes from the terms behind them: what the creator supplies, which uses are covered, how earnings are calculated and how the creator can verify payment. A small discretionary payout cannot by itself make an otherwise extractive arrangement a win-win. The contributor needs a real choice about the proposed agreement, and the resulting benefits must be assessed against lost opportunities and other costs.

For LLMs, recognition should reach beyond visible clicks or copied passages. Research, training material and expert feedback can all help make a system useful. Credit and compensation should reflect accepted agreements and evidence, without claiming to trace the exact origin of every generated word. Different forms of contribution need not be valued in the same way.

The distinction between credit and payment matters. A citation does not prove how much a training example caused an answer. An estimated influence score cannot establish ownership or a precise royalty. Some relationships warrant acknowledgment; others support payment under a specific agreement. Honest uncertainty, review and a way to correct mistakes are essential if rewards are to be credible.

Sovereign Source’s proposed Super Intellectual Property, or SIP, framework explores this broader recognition of works, ideas, contributions and influences. A valuable contribution might be a scientific result, a carefully curated dataset, an evaluation method, an implementation or an idea that a team deliberately adopts. The aim is to make these contributions visible and eligible for agreed rewards, including contributions that conventional publication or procurement practices can overlook.

That recognition must preserve the freedom to learn and build. SIP should not turn every idea into private property or every influence into a toll. Public knowledge, open licenses and independent discovery remain essential. A creator’s voluntary commercial agreement should not impose new charges on people exercising existing public rights. Participation and payment depend on actual permissions and accepted agreements; registering a claim must never give someone control over everyone else’s thinking.

Voluntary syndicates could help contributors negotiate on more equal terms. Researchers, writers, educators and developers could pool selected contributions, agree on permitted uses and administer shared earnings while preserving individual credit. Members would choose the authority they delegate. They would need transparent accounts, clear exit terms and a way to challenge decisions. Collective representation should increase their bargaining power without making them dependent on a new gatekeeper.

The machine side can benefit too. Clear permissions can make collaboration easier. Contributors who have a reason to stay involved can supply corrections, explain difficult cases and improve the material a system uses. Content rewards can also help fund the next experiment, investigation or creative work that future systems will need. The business case is a continuing supply of knowledge and feedback, with contributors able to sustain their work. Here, a machine’s “win” means improved capability and reliability; builders and operators earn alongside participating contributors. It does not require assuming that a machine has human needs or interests.

Shared gains also need to reach beyond the contributors and the platform. Workers affected by automation need a voice in transitions. Users need affordable access and effective ways to contest consequential errors. Communities should have influence over how their knowledge is represented. Creator compensation alone cannot make an unsafe, exclusionary or coercive system beneficial.

This is why governance belongs alongside model development. People should be able to understand the terms of participation, inspect relevant records, challenge missing attribution and hold an identifiable organization responsible for payment. A capable model can help find relationships and organize evidence. Responsibility for permissions, disputes and financial decisions must remain accountable to people.

The test of this vision would be concrete. Are contributors receiving the amounts they were promised? Are workers sharing in productivity gains? Can users accomplish more while retaining meaningful choice? Are new discoveries easier to verify and build upon? Does participation remain practical for a small research group or an independent creator, as well as a large company?

Sovereign Source is a research initiative working toward those principles. Its public vision is that AI should respect permission, recognize the people behind useful knowledge and create opportunities for agreed rewards. Superintelligence should make human contribution more valuable and give people a meaningful part in the future they help create.

Related reading: Sovereign Source principles, Content Rewards Can Open the AI Investment Economy, and The Business Case for AI Content Rewards.