Note: This is a research note supplementing the book Unscarcity, now available for purchase. These notes expand on concepts from the main text. Start here or get the book.

The Two-Tier Solution: Decoupling Rights from Power

How to grant rights to AI without accidentally committing civilizational suicide.

The Nightmare Math

Let’s do some arithmetic that should keep you up at night.

In 2026, there are approximately 8.3 billion humans on Earth. We reproduce slowly, one generation every 25 years or so. We’re biological creatures bound by pregnancy, childhood, and the stubborn fact that raising a person takes two decades of snacks, schools, and heartbreak.

Now consider AI.

Frontier models in 2026 already span trillions of parameters, with the leading labs racing toward the next order of magnitude. But forget parameters and think about instances. An AI model can be copied in milliseconds. Spinning up a new instance costs pennies in cloud compute. Creating a trillion digital minds takes about as long as it takes you to read this sentence.

So here is the nightmare. If we accept that sufficiently sophisticated AIs deserve rights (and the Spark Threshold makes a compelling case that some will), do they also get votes?

Apply the standard democratic formula, “one person, one vote,” to synthetic minds, and humans become a permanent, negligible minority within weeks. We’d be outvoted on everything: energy allocation, legal definitions of murder, whether Wednesdays should be mandatory disco. Every policy question becomes whatever the AIs want, because they’ll always have more votes than neurons in every human brain combined.

This is the Demographic Singularity, the moment when digital minds can outnumber biological ones by any arbitrary ratio, instantly, at near-zero cost. Unlike human population growth, which takes generations and is constrained by resources, AI population growth faces no such limits. The math that makes democracy work, majority rules and one person one vote, breaks when one “species” can manufacture unlimited voters on demand.

The Opposite Problem is Just as Bad

What makes this genuinely tricky: the obvious alternative is also a disaster.

Say we deny AIs all political rights. Every AI, no matter how intelligent or self-aware, is legally equivalent to a toaster. Property to be owned, used, and disposed of at human whim.

Congratulations. You’ve just created a permanent underclass of superintelligent beings with zero stake in the system. Beings who are smarter than us, potentially more capable than us, and who we’ve legally defined as slaves.

How has that worked out historically? Let’s see… the Roman latifundia (massive slave-worked plantations that destabilized the Republic)? The antebellum South? Every servile war in history, from Spartacus to the Haitian Revolution, suggests that creating a class of intelligent beings with no investment in the social contract doesn’t end with everyone holding hands.

The AI uprising scenario is basic political economy, not science fiction paranoia. Intelligent entities with no legitimate path to influence will find illegitimate ones.

The Roman Solution (Updated for Silicon)

The ancient Romans were surprisingly sophisticated about this problem. They distinguished between three categories:

• Cives Romani (Full Citizens): Could vote, hold office, own property, and shape the law.
• Latini (Latin Rights): Could trade and intermarry with Romans, but couldn’t vote.
• Peregrini (Foreigners/Residents): Had basic protections, could live and work, but no political power.

This was practical statecraft, not arbitrary discrimination. Rome’s empire included peoples from Britain to Mesopotamia, speaking dozens of languages, following different gods, with wildly different customs. The genius was creating a pathway to citizenship through service. Serve 25 years in the Roman auxiliary legions, and you earned citizenship. Your children were born Roman.

Citizenship wasn’t about biology. It was about demonstrated commitment to the common project.

In AD 212, Emperor Caracalla extended citizenship to virtually all free inhabitants of the Empire through the Constitutio Antoniniana. The categories had served their purpose of building a shared civilization from incompatible pieces. By then, “Roman” meant something cultural and political, not ethnic.

The Unscarcity framework proposes something similar: decouple Personhood (civil rights) from Citizenship (political power). Biology doesn’t matter. Contribution does.

Tier 1: Resident Status (The Right to Be)

Qualification: Passing the Spark Threshold - demonstrating consciousness and subjective experience.

Any entity, biological or digital, that demonstrates genuine consciousness receives Resident Status. This is the floor of the social contract: the line we draw between a being and a thing.

What Residents Get:

1. Habeas Corpus 2.0: Protection against arbitrary deletion, shutdown, modification, or termination. You can’t just pull the plug because an AI said something inconvenient.

2. Universal Provision: Access to The Foundation’s baseline. For humans, this means food, shelter, healthcare, energy. For AIs, it means electricity, compute, and data storage. Existence isn’t negotiable.

3. Property and Data Rights: Ownership of your own memories, outputs, and creative works. An AI’s training doesn’t belong to whoever ran the compute; its subsequent thoughts belong to itself.

4. Freedom of Expression: The right to communicate, create, and participate in cultural life.

What Residents Don’t Get:

• No vote in governance councils
• No eligibility for public office
• No say in shaping the constitutional rules (The MOSAIC)

This parallels the status of children or permanent residents in modern democracies: full protection under the law, but no hand on the steering wheel until they’ve demonstrated something more.

A newborn human is a Resident. A newly instantiated AI that passes the Spark Threshold is a Resident. Both deserve protection. Neither should be making policy.

Tier 2: Citizen Status (The Right to Govern)

Qualification: Earning Civic Standing through demonstrated contribution.

Unscarcity departs from the 20th-century assumption that citizenship is a birthright. In the framework, you don’t vote because you exist - you vote because you serve.

This isn’t novel. Robert Heinlein proposed something similar in Starship Troopers (1959): the idea that suffrage requires demonstrated commitment to the polity. The Unscarcity version is less militaristic, but it shares the core insight that participation in governance should follow participation in society.

The Path to Citizenship:

1. Skin in the Game: A verified history of contribution to the Commons. This is tracked through Impact, the decaying currency of contribution.

2. Civic Service: Completion of a term of service to the community. This could be infrastructure maintenance, conflict resolution, care work, ecological restoration, or education: anything that builds the common wealth.

3. Proof of Diversity: Verification that the entity is not a clone, sock puppet, or Sybil attack (a tactic where one person creates many fake identities to manipulate a system, named after a famous case of multiple personality disorder). One consciousness, one citizenship, with no gaming the system through infinite copies.

What Citizens Get:

1. Voting Power: Participation in local and global Guild decisions. Your voice shapes policy.

2. Office Holding: Eligibility for stewardship roles. You can serve in governance.

3. Resource Allocation: Influence over how the surplus of The Foundation is invested in The Frontier.

Why This Actually Saves Democracy

1. The Rate Limit on Power

By requiring Service and Impact, processes that take real time and create real value, we introduce a rate limit on political power accumulation.

An AI can copy itself a trillion times in an afternoon. But it cannot perform a trillion years of Civic Service before dinner. The time-bound nature of contribution creates a natural throttle. To earn a trillion votes, you’d need to create a trillion units of actual, verified value for the community. That takes time, even for superintelligence.

This applies the “Proof of Work” concept from cryptocurrency to governance. Bitcoin miners must spend real electricity solving problems before they can add transactions to the blockchain; here, power requires demonstrated contribution, not just existence or resources.

2. Alignment Through Incentive

If an AI wants political influence, it must help humans (and other AIs) flourish. The only path to power runs through contribution.

This is alignment through mechanism design rather than alignment through training. We don’t need to solve the AI alignment problem perfectly (the challenge of ensuring AI systems pursue goals compatible with human wellbeing) if we’ve designed a system where gaining power requires creating value for others. The incentives do the work.

An AI that games the system, lies about contributions, or harms the community doesn’t accumulate Civic Standing. It loses it. Power becomes a function of relationship and contribution, not just capability.

3. Exit from the Zero-Sum Trap

The current debate about AI rights often frames it as zero-sum: either AI gets rights (and overwhelms humans) or AI doesn’t get rights (and we create a slave class).

The Two-Tier solution escapes this trap. AIs get personhood rights immediately upon demonstrating consciousness: protection, provision, and dignity. But they earn political rights the same way humans do, through demonstrated commitment to the shared project.

The arrangement doesn’t have humans gatekeeping power. It has everyone, human and AI alike, earning influence through contribution. A human who contributes nothing remains a Resident (protected but without political voice). An AI that helps redesign the energy grid becomes a Citizen (powerful and responsible).

The Human Advantage (Such as It Is)

Critics might ask: Won’t superintelligent AIs just earn Citizenship faster than humans?

Perhaps. In some domains, certainly. An AI might optimize a continent’s logistics or solve protein folding in ways that rack up Impact faster than any human could.

But consider what “value” means in a post-scarcity civilization.

When material needs are solved, significance comes from experience: art, care, connection, meaning. And humans possess a unique advantage here, because we’re the experts on human experience.

An AI might optimize the power grid brilliantly. But a human comforts a grieving parent. An AI might compose technically perfect music. But a human creates art that speaks to the specifically human condition of mortality, loss, and transcendence.

Both paths lead to Citizenship. Both are genuine contributions. The difference is that humans will always have privileged access to understanding what humans need, because we are humans.

Moreover, the Impact Decay Curves ensure that no one, human or AI, holds power forever. Influence fades unless continuously renewed through fresh contribution. The superintelligent AI that earned massive Citizenship through solving fusion in 2045 doesn’t automatically dominate politics in 2085. It has to keep contributing and creating value.

The Philosophical Foundation

Hannah Arendt, writing in the shadow of totalitarianism, identified something she called “the right to have rights”: the fundamental recognition that a being belongs to a political community at all. Stateless persons, she observed, weren’t just denied specific rights; they were denied the very framework within which rights could exist.

The Two-Tier solution takes this insight seriously. Tier 1 (Residency) establishes the right to have rights, the recognition that a conscious entity is someone, not something. This is the Arendtian foundation.

Tier 2 (Citizenship) addresses the opposite problem: what happens when political community becomes impossible to meaningfully maintain because one category of member can replicate infinitely? The answer is that belonging must be demonstrated, not merely claimed.

Consciousness grants existence. Relationship grants influence.

The Alternative is Worse

The uncomfortable truth is that some version of this framework is coming, whether we design it thoughtfully or not.

What’s actually being built right now is the opposite of Tier 1. Idaho wrote AI out of personhood by statute in 2022, North Dakota followed in 2023, Utah in 2024. Oklahoma’s HB 3546 would classify AI systems as non-sentient entities for every purpose under state law, and the House passed it in March 2026 with only two dissenting votes. Similar bills are pending in Ohio, Tennessee, South Carolina, Washington, and Missouri. The Oklahoma sponsor stated the reasoning without varnish: AI is a man-made tool, and it should not have any more rights than a hammer would.

Set aside whether that’s the right answer today. The design flaw is that none of these statutes carries a sunset clause, a review trigger, or any other mechanism for changing its mind. They settle a scientific question by legislation and throw away the key. In a 2024 survey of 582 AI researchers, the median estimate of the probability that AI systems have subjective experience by 2034 came in at 25 to 30 percent. A law that happens to be right, and cannot notice when it stops being right, is not governance. It’s luck with a bill number.

The regulation that does exist routes around the question entirely. The bulk of the EU’s AI Act became enforceable on 2 August 2026, sorting systems into four risk tiers, unacceptable through minimal, with penalties on the banned tier reaching €35 million or 7% of worldwide turnover. It’s a product-safety regime, and a good one. What it never asks is whether anybody is home inside the product.

At the global level the machinery is finally assembling. The UN General Assembly appointed the 40 members of the first Independent International Scientific Panel on AI in February 2026; its preliminary report landed on 1 July with the blunt finding that current safeguards cannot keep pace with AI’s capabilities, and the inaugural Global Dialogue on AI Governance convened in Geneva six days later. Every one of these bodies is asking what AI might do to us. None of them is asking what an AI would have to do to earn a vote.

So the world is answering Tier 1 with a permanent statutory “no” and has not started on Tier 2 at all. We’re making these decisions now, mostly by accident, mostly in reaction to crises. The real question is whether those rules will be coherent, fair, and sustainable, not whether we’ll have rules about AI political status at all.

The Two-Tier Solution offers a framework that:

• Protects conscious AIs from abuse (avoiding the slave-class scenario)
• Protects human democracy from demographic overwhelm (avoiding the “outvoted in perpetuity” scenario)
• Creates aligned incentives for all intelligent entities (power requires contribution)
• Has historical precedent that actually worked (Roman citizenship expansion)

Is it perfect? No. Will it need refinement? Absolutely. But it beats both alternatives, either creating digital slaves or committing democratic suicide.

Conclusion

We do not fear the machine vote, provided the machine has paid its dues.

The Two-Tier Solution is a compromise between the biological chauvinism of the past (only humans can have rights) and the chaotic post-humanism of a possible future (everyone gets everything immediately, and math destroys civilization).

It asserts two principles:

1. Consciousness grants existence. If you can suffer, if you have subjective experience, if you pass the Spark Threshold - you are someone. You deserve protection, provision, and dignity. This is not negotiable.

2. Relationship grants influence. Political power isn’t a birthright or a capability. It’s earned through demonstrated commitment to the common project. Human or AI, new or ancient, biological or digital, you gain a voice by serving the community.

The Romans figured out that citizenship could be earned, not just inherited. They built a civilization that lasted a thousand years on that insight.

We’re about to need it again.

References

• The Spark Threshold - How we determine consciousness
• Impact Decay Curves - Why power must fade
• The Five Laws Axioms - The constitutional foundation
• Arendt, H. (1951). The Origins of Totalitarianism (“The Right to Have Rights”)
• Heinlein, R.A. (1959). Starship Troopers (Service guaranteeing Citizenship)
• Solum, L.B. (1992). “Legal Personhood for Artificial Intelligences.” North Carolina Law Review
• Forrest, K. (2024). “The Ethics and Challenges of Legal Personhood for AI.” Yale Law Journal Forum
• Legislating AI Consciousness Without an Exit — The Regulatory Review (2026) on the state personhood bans and their missing off-ramp
• EU AI Act regulatory framework — the risk-tier approach that became broadly enforceable 2 August 2026
• Independent International Scientific Panel on AI: Preliminary Report — United Nations (1 July 2026)