Overview of Constitution Breakdown #9: Alondra Nelson
This episode of 99% Invisible: The Breakdown of the Constitution covers Articles VI and VII of the U.S. Constitution, with a strong focus on the Supremacy Clause, federal preemption, and how those constitutional ideas shape modern debates over artificial intelligence regulation. Roman Mars and Elizabeth Jo first explain the historical purpose of the lesser-discussed constitutional clauses, then interview scholar and former White House OSTP director Dr. Alondra Nelson about the AI Bill of Rights, state vs. federal regulation, and why AI governance is becoming one of the most important constitutional policy fights of the moment.
Articles VI and VII: The Constitutional Basics
Article VII: Ratification
- Article VII is the Constitution’s ratification clause.
- It required nine states to ratify the document for it to take effect.
- That happened on June 21, 1788, when New Hampshire became the ninth state to ratify.
- The hosts note that this clause is not controversial today, but it was essential to making the Constitution legally valid.
Article VI: A Mix of Structural Rules
Article VI includes three major ideas:
- Debts Clause: The new United States would honor debts incurred under the Articles of Confederation and the Revolutionary era.
- Supremacy Clause: Federal law is the “supreme law of the land.”
- No Religious Test Clause: Public office cannot be conditioned on a religious test.
Why the Supremacy Clause Matters
Federal law overrides conflicting state law
- The Supremacy Clause exists to solve the problem of state and federal laws conflicting.
- Under the Articles of Confederation, there was no clear rule saying federal law came first.
- The Constitution fixes that by making federal law supreme over conflicting state law.
Preemption in practice
- The episode explains preemption as the practical extension of supremacy: when Congress passes a law, it can override state or local laws in that area.
- Courts often have to decide:
- Whether Congress intended to preempt state law
- Whether state and federal law can both be followed
- Whether the state law creates an obstacle or conflict
- The hosts describe preemption as one of the most frequently relevant constitutional doctrines in modern life, especially because states and the federal government regulate many of the same areas.
The No Religious Test Clause
- Article VI explicitly bars any religious test for public office.
- The episode traces this idea back to English legal traditions, where officials often had to swear loyalty to the Church of England or denounce Catholicism.
- The clause matters because it rejects that formal exclusion, even if religious representation in public life is still uneven.
- The hosts note that most modern legal disputes over religion now arise under the First Amendment, not this clause.
Dr. Alondra Nelson on AI and the Constitution
What counts as AI?
Nelson defines AI broadly as:
- Machine-based systems that use statistics and math to make inferences
- Systems that generate outputs such as:
- predictions
- recommendations
- decisions
- generated text, images, or sound
- Systems with varying degrees of:
- autonomy
- adaptability
- brittleness
She emphasizes that generative AI is only one part of the larger AI ecosystem, even though it gets most of the public attention.
Why AI should concern everyone
Nelson gives examples of AI’s benefits and harms:
- Potential benefits
- Reading medical scans and spotting early cancer
- Helping farmers detect crop disease
- Managing traffic flow and stoplights
- Supporting space exploration and simulation
- Potential harms
- Resume-screening systems that exclude qualified candidates
- Predictive policing systems that reinforce historical bias
- Municipal chatbots that give illegal or harmful advice
- Misuse in deepfakes, scams, and sexual abuse content
Her core point: AI affects jobs, housing, health care, education, policing, and civil rights, so it cannot be treated as a niche technical issue.
The AI Bill of Rights
Why it was created
Nelson describes the Biden administration’s Blueprint for an AI Bill of Rights as a civil-rights-style framework for governing powerful AI systems.
It emerged from:
- the pandemic
- the racial reckoning of 2020
- growing concerns about algorithmic harm
- broad public engagement with researchers, students, rabbis, advocates, and industry
The five principles
The blueprint’s five principles are:
- Safe and effective systems
- Protection from algorithmic discrimination
- Data privacy
- Notice and explanation
- Human alternatives, consideration, and fallback
What these principles mean
- AI systems should be tested before release.
- People should know when AI is making consequential decisions about them.
- Users should be able to get explanations when decisions matter.
- There should be a human option, especially in high-stakes areas like:
- housing
- health insurance
- employment
Nelson frames the AI Bill of Rights as a kind of civic infrastructure that gives ordinary people language for demanding fairer technology.
Biden vs. Trump: Two Very Different AI Philosophies
Biden administration approach
- Emphasized safety, ethics, civil rights, and guardrails
- Used executive authority to direct federal agencies
- Could not fully preempt state law because only Congress can legislate
- Asked agencies like Education and Labor to think about how AI affects schools and jobs
- Also addressed national security concerns and reporting requirements for advanced AI
Trump administration approach
- Reversed Biden’s executive order
- Shifted focus away from safety and ethics
- Prioritized accelerating AI development
- Called for federal preemption of state AI laws
- Has not yet gotten Congress to act
Why States Are Leading on AI Regulation
States are filling the vacuum
Nelson argues that because Congress has not passed meaningful AI legislation, states are doing the work instead.
Examples discussed:
- California: disclosure requirements for AI developers and AI use in police reports
- Colorado: an anti-discrimination AI law
- Texas: a more intent-focused approach
- New York: transparency and disclosure efforts
- Florida and Oklahoma: versions of an AI Bill of Rights
Why state regulation makes sense
Nelson’s argument:
- States are closer to the harms
- Local officials hear directly from affected constituents
- Different states can test different approaches
- A patchwork can create upward pressure for federal action
The episode also rejects the idea that this creates uniquely excessive burdens for companies, pointing out that businesses already navigate different state rules in areas like:
- insurance
- privacy
- employment
- biometric data
The Risk of Advanced AI and AGI
Nelson’s view of “AGI”
- She says she is not very invested in the label itself.
- She prefers the broader term advanced AI.
- Her concern is less about whether a system is called AGI and more about powerful systems with access to large data sets, autonomy, and broad decision-making capacity.
Why this matters
She raises concerns that advanced systems could:
- combine sensitive data across government and private systems
- intensify surveillance
- make autonomous decisions with little oversight
- create serious risks even before anything resembling “superintelligence” exists
Her key regulatory point:
- Companies do not need to be told what to invent, but they can be told what they are allowed to deploy.
- Stronger oversight could include:
- compute limits
- data access limits
- human approval before release
- emergency shutoff mechanisms
Thick Alignment: A Broader Model for AI Governance
What “thick alignment” means
Nelson expands the technical idea of AI alignment into a more social and political concept.
Instead of asking only:
- “Does the system technically work?”
She asks:
- “Does it work in context?”
- “Does it reflect community values?”
- “Who gets to define the goals?”
- “How do we account for different social needs and harms?”
Why it matters
- A system may be technically “aligned” and still be socially harmful.
- AI governance should consider:
- local values
- civil rights
- cultural context
- public accountability
She connects this to constitutional thinking: AI governance should not just be about engineering correctness, but about shared democratic values.
Main Takeaways
- Article VII made the Constitution legally effective through ratification by nine states.
- Article VI establishes important structural rules: federal debt obligations, the Supremacy Clause, and no religious tests for office.
- The Supremacy Clause is central to modern constitutional law because it resolves conflicts between federal and state law.
- AI regulation is now a major federalism issue: states are moving ahead while Congress remains largely inactive.
- Dr. Alondra Nelson’s AI Bill of Rights offers a rights-based framework for regulating AI around safety, discrimination, privacy, transparency, and human alternatives.
- The episode argues that state experimentation is valuable, especially when federal law is stalled.
- The broader concern is not just AI’s technical capacity, but its potential to reshape rights, labor, housing, education, and democracy itself.
Closing Note
The episode ends by suggesting that the AI debate is not just a technology story, but a constitutional and political one: who gets to regulate powerful systems, whose values count, and whether democratic institutions can keep pace with fast-moving technology.