AI and the American Future: Toward a Thriving Social Contract

A Public Learning Project of The People’s Commission on Technology and the American Future

An Editor’s Note from Moonshot Press

Moonshot Press is a human-led, AI-augmented publication. This article was written and edited by Shimon Waldfogel, MD, with artificial intelligence used to support research, questioning, synthesis, and editorial development. Shimon retains responsibility for the final judgments and published text.

AI collaborators used in developing this article: [list only the models actually used].

Coauthors with ChatGPT 5.6 Sol, Claude Opus 5

Learn more about how Moonshot Press uses artificial intelligence.

Imagine a billing specialist in a mid-sized healthcare company. For fifteen years, she has known the intricate rules that determine what insurers will pay, how to correct an error before it becomes a denial, and how to help a patient understand a bill that makes no sense to them.

Her work is not glamorous. Much of it is repetitive. Some of it is frustrating. But she is good at it. People come to her when a problem seems impossible. Her job gives shape to her week, a group of colleagues who know her name, a reason to leave the house, money to help support her family, and the quiet satisfaction of being useful to people who need help.

Then the company installs an AI system that can process most claims, identify likely errors, and generate patient explanations in seconds.

At first, the change feels like relief. The system takes over tedious tasks. It reduces backlogs. It may even spare patients some of the confusion and delay that made the old system so difficult.

But after several months, the company decides it needs fewer specialists. The remaining workers are assigned to supervise the system, manage exceptions, and meet higher productivity targets. The work becomes less predictable. Her judgment matters less. Her role narrows. She begins to wonder whether the competence she spent years developing still has a place.

Nothing in this story requires us to be against technology. The new system may genuinely improve care. It may reduce costs. It may free some workers to focus on the human problems machines cannot resolve.

But it raises a question that lies beneath almost every public argument about artificial intelligence:

If a machine can do part of a person’s job, what exactly is at risk—and what, if anything, has been gained?

That question is larger than employment statistics. It is larger than whether a particular company hires or lays off workers this year. It reaches into how people understand their place in the world, how communities remain connected, and what democratic society owes its members when the conditions of work change.

Before we can decide what AI should do, we need to ask what work has been doing for us.

More than a paycheck

For most adults, paid employment has been the institution through which many essential goods are bundled together.

A job provides income. That is obvious, and it is indispensable. Income determines whether people can afford housing, food, healthcare, childcare, transportation, and the ordinary security required to plan a life.

But work has often provided much more.

It gives the day a shape. It creates routines and expectations. It asks people to develop skill, exercise judgment, solve problems, cooperate with others, and become reliable in the eyes of colleagues and customers. It can connect effort to result: I learned something difficult; I did it well; someone depended on me; my work made a difference.

Work can also confer standing. When people meet, one of the first questions they often ask is, “What do you do?” The question may be clumsy, but it reflects a deep habit of social life. Occupation has become one way people explain who they are, what they know, and how they contribute.

That arrangement has never been fair or complete. Many jobs are monotonous, unsafe, poorly paid, or degrading. Many forms of vital work—raising children, caring for elders, maintaining a household, volunteering, organizing a neighborhood, supporting a friend through illness—are unpaid or underrecognized. Many people have been excluded from stable, dignified employment by discrimination, disability, immigration status, geography, or the simple accident of being born into a community with too little opportunity.

Still, for millions of Americans, paid work has been the principal place where income, identity, structure, relationships, and contribution come together.

When that bundle breaks apart, a person can lose much more than a wage.

Work, employment, contribution, and human worth

The distinction among these terms matters.

Human worth is not earned through a job. A person who is unemployed, disabled, retired, caring for family, studying, grieving, or unable to work has no less dignity than a person with a prestigious title or a high salary.

Work is broader than paid employment. It includes care, service, learning, craft, creative effort, civic responsibility, and the daily labor of sustaining other people and communities.

Employment is a particular arrangement: an institution that exchanges labor for income and, at its best, also supplies relationships, recognition, benefits, structure, and a path of development.

Contribution is the experience of being needed by others. It is the sense that something one does matters beyond oneself.

A healthy society should not make a person’s right to dignity depend on whether an employer needs them. But neither should it dismiss the real losses people experience when work disappears or becomes hollowed out.

The public conversation sometimes offers a false choice. Either we preserve every existing job, or we accept that technology will make many jobs unnecessary and assume people will be fine if they receive an income.

Neither answer is adequate.

Some jobs should change. Work that is dangerous, degrading, or needlessly tedious should not be preserved simply because it is familiar. If AI can reduce paperwork that keeps nurses from patients, lessen repetitive administrative work, make a small business more viable, or help a teacher adapt instruction to a child’s needs, those are gains worth welcoming.

But income alone cannot automatically replace everything a job provided. A monthly payment may help a family survive. It does not by itself create a community of practice, a sense of mastery, a reason to get up in the morning, or confidence that one has something valued to offer.

The question is not whether every old job must remain. The question is whether people will retain meaningful ways to belong, contribute, learn, care, and exercise agency in the world around them.

The conditions of coherence

The medical sociologist Aaron Antonovsky devoted his work to a deceptively simple question: not only what makes people ill, but what helps them remain healthy under conditions of stress.

His answer centered on what he called a Sense of Coherence. People are more resilient when life is experienced as:

Comprehensible — The world makes enough sense that I can understand what is happening and why.

Manageable — I have, or can gain access to, the resources needed to meet life’s demands.

Meaningful — My effort is worth making; I have reason to invest myself in the challenges before me.

These are not luxuries. They are conditions that help people navigate uncertainty without becoming overwhelmed by it.

Good work can support all three.

A worker whose effort leads to visible results may experience the world as more comprehensible. A job can provide wages, training, colleagues, credentials, and routines that make life more manageable. A person whose skills help others may experience meaningfulness: I am capable. I am useful. My contribution matters.

But work can also undermine coherence. A job that is insecure, arbitrary, dangerous, humiliating, or impossible to understand does not necessarily create health. Neither does a workplace that treats people only as costs to be reduced or data points to be managed.

This is why the future of work cannot be judged simply by counting how many jobs exist. We need to ask what kinds of work are being created, transformed, or eliminated—and what happens to the people whose lives were organized around them.

AI: liberation, disruption, or both?

Artificial intelligence is not one thing, and its effects will not be one thing.

In one workplace, AI may reduce drudgery. A physician may spend less time documenting a visit and more time listening to a patient. A public-interest lawyer may analyze a large body of records more quickly and devote more attention to strategy and human judgment. A small business owner may use AI to translate materials, draft a marketing plan, organize accounts, or provide better customer service without hiring a large back-office staff.

In these cases, AI can operate as a tool of augmentation. It may increase people’s capacity to do work they consider more skilled, more relational, or more meaningful.

In another workplace, the same technology may remove entry-level roles, concentrate authority in fewer hands, weaken a worker’s discretion, intensify surveillance, or turn a skilled occupation into a narrow task of monitoring a system someone else designed. In those cases, AI may reduce costs and increase output while diminishing people’s ability to exercise judgment, build mastery, or see themselves as contributors.

The difference is not merely technical. It is organizational and political.

Who decides how the technology is introduced? Do workers have a voice? Are productivity gains shared? Does automation remove a burden while preserving the worker’s role, or does it remove the worker’s role while preserving the burden for someone else? Are people given time, training, security, and real choices—or simply told to adapt?

No one can yet answer these questions for every occupation. Predictions about the future of work vary widely. Some analysts expect AI to create new categories of work and raise productivity across the economy. Others expect rapid displacement, particularly in administrative, clerical, and professional roles once considered protected from automation. Both possibilities deserve serious attention.

What is clear is that the outcome will not be determined by AI alone.

What an adequate transition must protect

The familiar response to technological disruption is often one word: retraining.

Learning matters. People should have repeated opportunities to acquire new knowledge, new tools, and new forms of competence throughout life. A society facing rapid technological change cannot rely on a single credential earned in youth to sustain a career for decades.

But retraining is not a complete answer.

It assumes that new jobs will exist in sufficient number, that people can afford the period of transition, that training will lead to genuine opportunity, and that a person’s needs can be reduced to acquiring another marketable skill. Sometimes those assumptions will be true. Sometimes they will not.

A more adequate response must ask at least six questions.

Will people have economic security?
Can they pay bills, care for their families, maintain housing, and weather a transition without catastrophe?

Will people have access to learning and adaptation?
Can they gain useful skills, understand changing technology, and move into work or contribution that is genuinely available?

Will people retain voice and dignity at work?
Are workers consulted when AI changes their roles? Can they question unfair systems? Do they share in the gains their experience and labor help make possible?

Will communities remain capable of supporting one another?
When a local employer changes, contracts, or disappears, are there institutions—schools, libraries, community colleges, unions, faith communities, health systems, civic associations, local governments—ready to help people navigate the change?

Will people have meaningful ways to contribute?
If traditional employment becomes less stable or less central, what institutions will recognize caregiving, service, mentorship, creation, civic work, and other forms of needed contribution?

Will citizens remain able to participate in self-government?
Economic insecurity narrows the time, attention, confidence, and social connection people need to participate in public life. A democratic society cannot afford an economy that leaves large numbers of citizens too precarious, exhausted, or isolated to exercise their voice.

These questions do not yet tell us what policy to adopt. They tell us what a serious policy must be capable of seeing.

An open question for citizens

There is a genuine disagreement here.

One view holds that the best response to AI is to embrace innovation, allow old roles to change, and focus on creating new opportunity. History offers reasons for this optimism. Previous technological transformations disrupted lives but also produced new industries, new professions, greater productivity, and improvements in material well-being.

Another view holds that the current transition may move too quickly, concentrate too much power, and reach too deeply into cognitive and professional work for existing institutions to absorb it. History also offers reasons for this concern. Technological gains have often been distributed unevenly, and communities have frequently borne the costs long before they received any benefits.

Both views contain truths.

The task of democratic citizenship is not to choose between optimism and fear. It is to refuse the false comfort of either one.

We should welcome technologies that expand human capability, reduce needless suffering, and make more room for care, creativity, learning, and contribution. We should also insist that no person be treated as disposable because a machine can perform some of the tasks they once performed.

The question is not whether the future will contain less work, more work, different work, or new forms of contribution we cannot yet imagine.

The question is whether we will build a society in which people remain secure, capable, connected, and needed.

When technology changes the role of human labor, what do we owe one another—not merely so people can survive, but so they can continue to belong, contribute, and flourish?

The next chapter turns to the technology itself: what artificial intelligence can actually do, what it cannot do, and why understanding that difference is essential to shaping the future wisely.

Discussion questions

1. Which functions of work—income, structure, mastery, belonging, contribution, or civic capacity—are most difficult to replace when a job disappears?

2. When AI removes repetitive or burdensome tasks, what should employers and workers do to make sure the gain becomes more meaningful human work rather than simply fewer human workers?

3. If paid employment becomes less stable or less central to adult life, what new institutions might help people experience dignity, security, and the sense of being needed?

Selected foundations for further reading

Aaron Antonovsky, Unraveling the Mystery of Health; Hannah Arendt, The Human Condition; Adam Smith, The Theory of Moral Sentiments and The Wealth of Nations; David Autor, “Work of the Past, Work of the Future”; Daron Acemoglu and Simon Johnson, Power and Progress.

An Editor’s Note from Moonshot Press

Moonshot Press is a human-led, AI-augmented publication. This article was written and edited by Shimon Waldfogel, MD, with artificial intelligence used to support research, questioning, synthesis, and editorial development. Shimon retains responsibility for the final judgments and published text.

AI collaborators used in developing this article: ChatGPT 5.6 Sol Claude Opus 5 High

Learn more about how Moonshot Press uses artificial intelligence.

An orientation to the Civic Curriculum for the AI Age

Consider a high school senior trying to decide what to study.

She is good at writing and good at math, and people who mean well—but cannot possibly know whether they are right—have told her to avoid anything a machine may soon be able to do. She asks what that leaves. Her guidance counselor does not have a reliable answer. Neither do the labor economists, the executives selling the systems, or the researchers building them.

She is being asked to make a consequential decision under genuine uncertainty, and the honest response—nobody knows—is not much comfort when the application deadline is in November.

Millions of people are in some version of her position. Not only students choosing a field, but a paralegal watching first-year work quietly disappear, a school board deciding what to buy, a county weighing a data center against its water supply, a physician deciding how much clinical judgment can safely be delegated to software. These decisions are being made right now, in laboratories and boardrooms and agencies and school districts and living rooms. They will shape how people work, how children learn, how healthcare is delivered, how information travels, and how power is distributed.

And most of us are being invited into that future as consumers, users, patients, and employees—not as citizens with any say in what the technology is for.

The Civic Curriculum begins from a different premise: citizens should help shape the technological future they will be asked to inhabit.

We do not know what is going to happen

Public conversation about AI rewards confidence. One camp anticipates extraordinary abundance—productivity growth, scientific breakthroughs, personalized education, liberation from tedious work. Another warns of mass unemployment, surveillance, manipulation, concentrated power, and the loss of meaningful human control. Both accounts deserve serious examination. Neither should be mistaken for a forecast.

Here is what we actually know. AI capabilities have advanced rapidly. Enormous sums are being invested. Machines can now assist with, and sometimes perform, work involving writing, coding, analysis, design, research, and clinical decision support that recently required trained human beings.

Here is what we do not know. We do not know how quickly technical capability converts into dependable economic value. We do not know how many occupations will vanish, how many will change, or how many new ones will appear. We do not know whether productivity gains will reach ordinary households or stay with those who own the systems. We do not know whether organizations will find these tools more expensive and unreliable than promised. And we do not know whether our institutions will adapt faster than the technology or considerably slower.

We do not know. That may be the most important fact with which to begin.

This is not a hedge, and it is not a reason for passivity. A society that waits until the consequences are unmistakable has waited too long. A society that reorganizes itself around predictions that turn out to be wrong does real damage of a different kind. The task is not to pick the most dramatic forecast and build around it. The task is to prepare for several plausible futures, watch the evidence as it comes in, and build institutions capable of changing course.

Five futures worth preparing for

Chapter 4 examines these in depth. For now, the map:

The AI Slowdown. AI proves useful but far less transformative than promised. The disruption comes from the expectations themselves—abandoned projects, failed companies, communities left holding stranded investments in infrastructure built for growth that never arrived.

The Assistive Economy. AI expands human capability without broadly replacing people. Doctors, teachers, tradespeople, and small-business owners become more capable. The central question becomes who receives the tools—because a technology that reaches only the well-resourced raises productivity while widening the gap.

Uneven Transformation. AI creates real value but reorganizes work faster than people and institutions can absorb. Entry-level pathways disappear in professions where machines now perform the work through which beginners once learned the craft. The aggregate economy looks healthy while particular people, occupations, and regions are hollowed out.

Extraction Without Abundance. The technology works, the gains are real, and they are captured. Output rises, wages do not, ownership concentrates, and the promised broad prosperity never materializes for most people.

Abundance Without Employment. The most radical scenario, and the one to hold most loosely. Machines become capable enough that human labor is much less necessary for production. Society grows wealthy while needing fewer people to produce that wealth—at which point income turns out to be the easier problem, and the harder one is what work was doing besides paying us.

These are scenarios, not predictions. Reality may combine them, or move between them, or land somewhere none of them anticipated. Their purpose is to widen the field of vision and to ask what citizens and institutions would need to do if each began to emerge.

The question underneath

I spent my career as a psychiatrist, and it shapes how I read all five of these.

Prolonged unemployment, social disconnection, and loss of purpose have been associated with serious consequences for mental and physical health—including depression, substance use, cardiovascular illness, and premature death. We watched a version of this move through deindustrializing towns a generation ago and eventually gave it a name.

Whether that pattern follows AI into offices, hospitals, classrooms, and law firms is not established. I think the risk is serious enough to examine before the evidence is conclusive rather than after. Almost nothing in the current policy conversation is built to catch it, because that conversation is organized around replacing income rather than restoring meaning.

There is a useful way to think about this, and it runs through the whole Curriculum. The medical sociologist Aaron Antonovsky asked an unusual question. Not what makes people sick—the question medicine had been asking—but what keeps people well, given that stress and hardship are universal? His answer centered on whether life is comprehensible, manageable, and meaningful: whether I can make sense of what is happening, whether I have the resources to meet it, and whether my effort is worth making.

Turn that question toward technology and the policy conversation reframes itself. Not how do we cushion the harms, but does this system generate the conditions in which people can be healthy? We usually ask whether an AI system is accurate, efficient, safe, and profitable. Those are the right questions and they are not sufficient. Alongside them: Can people understand what it is doing to them and contest an error? Does it strengthen human capability or create hidden dependence? Who receives the benefits, and who carries the burdens?

That standard is developed in Chapter 9. It is worth naming now, because it explains why this Curriculum is organized the way it is.

What this is, and how it will work

This is not a technical course, and it will not teach readers how to build AI systems. It is a shared framework for understanding what AI may change, deliberating about the choices before us, and identifying meaningful ways to act.

A few commitments:

I will not assume transformation is inevitable. Claims about what AI will do are claims to examine, not facts to inherit.

I will present the strongest competing arguments, evaluated on their evidence, assumptions, and consequences—including the ones I find least congenial.

I will distinguish evidence from inference from prediction. What is technically possible is not necessarily economically practical, institutionally adopted, or socially desirable.

I will preserve distinctions that matter. A task is not a job. A job is not an occupation. Employment is not the only form of contribution. Productivity is not human worth.

I will revise in public. This is a living inquiry, and I will get things wrong.

The Curriculum is the learning half. The People’s Commission on Technology and the American Future is the other half—a method for moving from understanding to deliberation to action. It is not a panel of experts speaking for everyone else. It is a structure for citizens to examine evidence, contribute what they know from their own lives, and make their conclusions visible to the people who hold power.

One helps us understand the choices. The other helps us exercise citizenship within them.

What is coming

This Orientation begins a twelve-part public-learning journey: ten chapters followed by an epilogue, published weekly through late October.

We begin with the human question—what work has been doing for us besides paying us. Then the technology itself: what these systems actually are, what they can and cannot do. Then history, and where it stops being a reliable guide. Then the five futures in depth. Then power: who gains, who loses, who decides, and who governs any of this. Then choices—the real policy options and the honest arguments against each. Then information and democratic power, the salutogenic standard, and finally what Americans should be able to expect from one another in an age of intelligent machines.

Each chapter includes questions for reflection and for group deliberation. They are meant to be argued with.

The choice before us

Whatever future arrives, one principle runs through all five scenarios: technology changes what is possible; it does not decide what is just.

Those decisions—who gets access, who bears risk, who captures gains, whose rights are protected—will be made by people. Some in laboratories and boardrooms, others by legislatures, courts, agencies, schools, hospitals, states, cities, and citizens. They are not merely technical choices, and they are not merely economic ones. They are choices about the kind of society we intend to become.

The future of artificial intelligence will not be determined by citizens alone. But it should not be determined without us.

The Curriculum begins with learning, but its destination is participation.

Chapter 1 arrives Monday, August 24: What Is Work For? It begins with a woman who has spent fifteen years becoming very good at a job that a new system can now mostly do—and asks what exactly is lost when that happens, beyond the paycheck.

If you want to be part of the deliberation rather than only the readership, join the Moonshot and tell me what this transition makes you hopeful about, worried about, or curious about. Your answers will shape what I investigate.

I am glad you are here.

Shimon Waldfogel, MD Founder and Publisher, Moonshot Press

Join the Commission

Share

Subscribe now

The companies building artificial intelligence are spending more than $660 billion this year on the bet that AI will transform the economy. The economists studying that bet disagree with each other by an order of magnitude about what will actually happen. The workers and communities in the path of the transformation have been given awareness of the disruption but almost no tools for evaluating the range of outcomes — or for holding their elected officials accountable to preparing for each of them.

This brief presents four plausible scenarios for the AI transition, describes what each means for workers and citizens, and provides specific questions that the People’s Council will pose to policymakers under each scenario. No one knows which scenario will unfold. All four require a democratic response that does not yet exist.

Scenario One: AI Delivers — The Transformation Is Real

What happens

AI achieves the productivity gains its proponents project. The investment proves justified. New industries and new categories of work emerge. The economy grows substantially — perhaps by the six to nine percent of GDP that optimistic projections envision. The technology becomes as foundational as electricity or the internet: disruptive in the transition, transformative in the outcome.

What it means for workers and citizens

Even in the best case, the transition is severe for those directly affected. Tens of millions of workers whose jobs are reorganized or eliminated face a period — years, not months — of displacement, retraining, and identity reconstruction. Historical precedent is instructive: electricity was invented in the 1880s but did not produce measurable aggregate productivity gains until the 1920s. The workers displaced during those four decades did not benefit from the eventual transformation. They bore the costs of a transition whose gains accrued to the next generation.

Under this scenario, the economy generates enough new wealth to fund generous transition programs — retraining, income support, portable benefits, community investment. But that wealth does not distribute itself. Every prior technology revolution that increased aggregate prosperity also concentrated that prosperity in the hands of technology owners, unless democratic institutions intervened. The Gilded Age was extraordinarily productive. It was also extraordinarily unequal. The question under Scenario One is not whether AI creates value. It is whether the institutions exist to ensure the value is shared.

What you would experience: If you work in a field directly affected by AI — knowledge work, professional services, creative industries, customer service, financial analysis, software development — you face a period of significant disruption even in the best case. Your job may not disappear, but it will change in ways that require new skills, new professional identities, and new relationships with AI tools. The transition may be manageable if your employer invests in genuine onboarding, if your community provides support infrastructure, and if the policy environment includes funded retraining, income bridges, and portable benefits. Without those supports, even a broadly positive transformation produces concentrated suffering among the people who happen to be in the wrong occupation at the wrong time.

Questions for policymakers

• If AI generates the productivity gains its proponents project, what specific mechanism ensures those gains are broadly distributed rather than captured by technology owners and shareholders? The tax code, the labor law framework, and the benefit structure were designed for an economy in which productivity gains flowed primarily through wages. If AI breaks that connection — producing more with fewer workers — what replaces it?

• What is your plan for funding transition programs at a scale commensurate with the disruption? The AI Workforce PREPARE Act authorizes six million dollars over five years. The companies driving the transition are spending $660 billion in a single year. What ratio between private investment in displacement and public investment in transition do you consider adequate?

• Do you support mandatory reporting on AI-driven workforce changes? Without data on which jobs are being eliminated, which are being created, and which communities are bearing the costs, every other policy response is guesswork. The Warner-Hawley AI-Related Job Impacts Clarity Act would require this reporting. What is your position?

Scenario Two: AI Delivers Partially — Transformative in Some Sectors, Disappointing in Others

What happens

AI produces genuine productivity gains in specific applications — software development, customer service, certain knowledge-work tasks, medical documentation — but falls short of the broad economic transformation that the investment levels assume. Five to thirteen percent of firms achieve significant returns. The rest experiment, pilot, and see modest or no measurable impact. The investment cycle does not crash but corrects: spending moderates, some AI startups fail, the surviving companies consolidate. The technology settles into the economy the way the internet did after the dot-com correction — transformative over two decades, but not overnight, and not evenly.

What it means for workers and citizens

This is, in many ways, the hardest scenario to navigate. The disruption is real but concentrated. Workers in the sectors where AI works well — call centers, routine document production, code generation, data analysis — face genuine displacement. Workers in sectors where AI underperforms — complex healthcare, education, skilled trades, relational professions — face augmentation rather than replacement, but also the anxiety of not knowing whether their sector is next.

The critical challenge under Scenario Two is that the economy has not grown enough to make generous transition programs economically painless. The aggregate gains are real but modest. The political case for large-scale public investment in displaced workers is weaker than under Scenario One, because the transformation feels less dramatic and the benefits are less visible. Yet the workers who are displaced face the same identity disruption, the same community consequences, and the same coherence crisis as under any scenario. Their suffering is not reduced by the fact that it is concentrated rather than universal.

What you would experience: The effects depend heavily on your sector and your employer’s choices. If you work in a sector where AI has proved effective, you may face the same displacement pressures as under Scenario One — but with less public sympathy and fewer resources, because the national conversation has moved on to debating whether AI was “overhyped.” If you work in a sector where AI has disappointed, you may feel relief but also uncertainty: the technology is still improving, the investment is still flowing, and the question of whether your field is next remains open. The most common experience under Scenario Two is not catastrophic job loss but a slow erosion of the tasks that defined your professional identity — a gradual shift from creator to supervisor of AI outputs, accompanied by the cognitive overload and identity disruption that the BCG research on “AI brain fry” has documented.

Questions for policymakers

• How will you ensure that workers in AI-disrupted sectors receive adequate transition support even if the overall economic impact of AI is modest? The political temptation under Scenario Two is to treat AI displacement as a normal market adjustment. The evidence from deindustrialization shows that concentrated displacement produces the same human consequences regardless of whether it registers in aggregate economic statistics. What is your commitment to sector-specific transition programs?

• What standards will you support for AI implementation in workplaces? The research shows that the design of AI deployment — not AI itself — determines whether workers experience augmentation or replacement, cognitive support or brain fry. Do you support requirements for worker voice in AI implementation decisions, cognitive load limits, and deskilling risk assessment in high-stakes domains?

• How will you address the gap between AI hype and AI reality in policymaking? Under Scenario Two, the danger is that policy is designed for a transformation that doesn’t fully materialize, or — more likely — that the absence of a dramatic transformation is used as an excuse for inaction. What evidence will you use to determine the appropriate scale of response, and how will you update that assessment as the technology evolves?

Scenario Three: The AI Bubble Bursts

What happens

The gap between AI infrastructure spending and AI-generated revenue proves unsustainable. A correction occurs — potentially gradual (spending moderates, valuations decline, marginal companies fail) or potentially sharp (a market event triggers rapid repricing of AI assets, cascading through the technology sector and into the broader economy). Historical parallels include the dot-com bust of 2000–2001, when the Nasdaq lost seventy-eight percent of its value, and the telecom bubble of the same era, when ninety-five percent of installed fiber optic cable went unused and the sector lost over two trillion dollars in market value.

What it means for workers and citizens

The critical finding from historical precedent is that displacement does not reverse when the investment thesis fails. After the dot-com crash, Silicon Valley took approximately sixteen years to recover to its prior employment levels. After the 2008 financial crisis, workers who lost full-time jobs experienced the worst re-employment outcomes in a generation: only thirty-five to forty percent regained full-time work within a year, and their earnings remained ten percent lower even a decade later.

Under Scenario Three, three categories of workers are affected simultaneously. First, the workers who were displaced during the AI expansion — whose jobs were automated or restructured during the boom — do not get those jobs back when the bubble bursts. The restructuring was real even if the investment thesis was not. Second, workers in the AI industry itself — the engineers, the data scientists, the infrastructure builders — face layoffs as spending contracts. Third, workers in the communities that oriented their economic strategies around AI infrastructure — the data center construction workers, the support services, the local businesses that served an expanded workforce — face the economic consequences of stranded assets. Data centers employ an average of 1,688 workers during construction but only 157 permanently. Communities built around construction-phase employment face a sharp contraction when building stops.

What you would experience: If you were displaced by AI during the boom and are still seeking re-employment when the correction arrives, you face a labor market that has worsened, not improved. If you work in AI or AI-adjacent industries, you face the same displacement cycle — but without the narrative of inevitable technological progress that justified the disruption. If you live in a community that courted data center development as an economic strategy, you may find that the infrastructure built during the boom has a useful life of three to five years, is poorly suited for alternative uses, and generates far fewer permanent jobs than the construction phase promised. The psychological dimension is also distinct: under Scenario Three, the disruption feels not only painful but pointless. The sacrifice was supposed to be in service of a transformation. If the transformation doesn’t come, the loss of identity and purpose is compounded by the sense that it was for nothing.

Questions for policymakers

• What contingency planning exists for an AI investment correction? Deutsche Bank has estimated that AI infrastructure spending contributes roughly 0.8 percentage points of total U.S. GDP growth. If that spending contracts sharply, what is the macroeconomic impact, and what countercyclical mechanisms are in place?

• What protections exist for communities that have oriented their economic development around AI infrastructure? Communities in Northern Virginia, central Ohio, and other data center corridors have made long-term commitments — tax incentives, grid infrastructure, land use changes — on the assumption of sustained AI investment. If that investment contracts, who bears the cost? Have those communities been provided with honest assessments of the risks alongside the promised benefits?

• What is your position on severance and transition requirements for AI-driven layoffs? Under current law, companies that restructure around AI and then retrench when the investment thesis fails face no obligation to the workers displaced in either direction. Do you support requirements for advance notice, severance, continued benefits, and retraining funding for workers displaced in AI-related restructuring?

• How will you address the “double hit” faced by workers displaced during the boom who remain unemployed during the bust? These workers — displaced by AI automation that proved sufficient to eliminate their positions but insufficient to generate the economic growth that was supposed to create new opportunities — represent the population most at risk for the long-term scarring effects documented in prior economic crises.

Scenario Four: The Worst of Both Worlds

What happens

AI proves capable enough to displace workers but not productive enough to generate the economic abundance that would offset that displacement. The MIT Nobel laureate Daron Acemoglu calls this “so-so automation” — technology that improves corporate margins without meaningfully improving overall productivity or creating new categories of employment. The Nobel laureate Joseph Stiglitz described this scenario in March 2026 as a situation in which “we do not have the macro or micro framework for managing that kind of displacement.”

A formal economic model published by researchers at the University of Pennsylvania and Boston University in March 2026 demonstrates why this outcome is not merely possible but structurally incentivized. In competitive markets, each firm captures the full cost savings of replacing workers with AI but bears only a fraction of the demand destruction that displaced workers represent. The result is a Prisoner’s Dilemma: every firm automates because its competitors are automating, and the collective result is that they destroy the consumer base they all depend on. The researchers tested the most commonly proposed solutions — wage adjustments, worker ownership stakes, universal basic income, retraining — and found that none of them eliminates the structural incentive to over-automate.

What it means for workers and citizens

Scenario Four combines the displacement of Scenarios One and Two with the economic weakness of Scenario Three. Workers lose jobs to AI automation, but the economy does not generate enough new activity to reabsorb them. Consumer demand weakens as displaced workers reduce spending. Companies respond to weakening demand by automating further — cutting costs to maintain margins — which displaces more workers, which weakens demand further. The cycle is self-reinforcing.

The historical parallel is the agricultural displacement of the early twentieth century. Mechanization dramatically increased farm productivity, but the displaced farmers had nowhere to go — the industrial economy had not yet grown large enough to absorb them. The result was two decades of rural poverty, mass migration, and social instability that was not resolved until World War II government spending created the demand that the private economy had not. The AI version of this pattern would play out among the professional and knowledge-working classes rather than rural farmers — but the structural logic is identical.

What you would experience: Under Scenario Four, the ground shifts beneath you in ways that feel individually random but are systemically connected. Your company automates a set of tasks and reduces headcount. You search for a new position and find that similar roles across the industry have been restructured. You pursue retraining, but the fields you retrain for are also being automated — the technology is improving faster than the retraining pipeline can adapt. The consumer economy weakens because millions of workers in your situation are spending less. Your community’s tax base contracts. The civic institutions that might have supported your transition — the library, the community college, the mental health services — face budget cuts. You are told that AI is creating new kinds of work, but the new jobs require skills you don’t have, are located in places you don’t live, or pay less than the position you lost. The anxiety is compounded by the sense that no one in a position of authority has a plan adequate to what you are experiencing.

Questions for policymakers

• Do you support an automation tax — a Pigouvian tax on AI-driven labor replacement — as a mechanism for aligning the private incentive to automate with the social cost of displacement? The Falk-Tsoukalas research demonstrates that voluntary measures, market corrections, and conventional policy tools cannot eliminate the structural incentive to over-automate. A tax on automation that is calibrated to the social cost of displacement — and whose revenue funds transition programs — is the only mechanism their model identifies as effective. What is your position?

• What is your plan for maintaining consumer demand if AI displacement reduces aggregate purchasing power? The consumption paradox — the structural contradiction of an economy that hollows out the purchasing power of its own customer base — is not a theoretical concern. It is the mechanism through which Scenario Four becomes self-reinforcing. What specific fiscal, monetary, or structural policies would you deploy to prevent the demand-destruction spiral?

• Do you support the creation of democratic advisory bodies on AI and the workforce that include workers, labor economists, community health practitioners, and civic leaders — not only technology executives and investors? The current Presidential Council of Advisors on Science and Technology includes twelve technology company executives and no labor representatives. Under Scenario Four, the governance of the AI transition cannot be entrusted to the industry driving the disruption. What structural changes to AI governance do you support?

• What is your standard for evaluating whether the AI transition is being managed adequately? The People’s Council applies the salutogenic standard: did we restore the conditions under which people can experience their working lives as comprehensible, manageable, and meaningful? That standard measures not only income replacement but identity reconstruction, community restoration, and democratic participation. What standard do you apply, and how will you report publicly on whether it is being met?

What All Four Scenarios Have in Common

Across every scenario, five findings hold:

Displacement does not reverse. Whether AI delivers or disappoints, workers who have been displaced, organizations that have restructured, and communities that have reoriented their economies do not return to their prior state. The restructuring is permanent; only the promised benefits are uncertain.

The existing policy infrastructure is inadequate. No scenario is well served by four pages of federal workforce recommendations, six million dollars in transition funding, or a governance structure composed exclusively of the industry driving the disruption.

The human consequences extend beyond income. Professional identity, community belonging, daily structure, purpose, and the sense that one’s life is comprehensible, manageable, and meaningful — these are disrupted under every scenario, and they are not restored by income replacement alone.

The asymmetry of consequences demands preparation. If we prepare aggressively and the optimists are right, the cost is modest: we will have built civic institutions that strengthen democratic life regardless. If we fail to prepare and the pessimists are right, the cost is measured in human lives and democratic capacity.

Workers and citizens must have a structural voice. Under no scenario is it adequate for the governance of the AI transition to be conducted by the companies driving it, advised by councils they populate, and covered by media platforms they increasingly own. Democratic voice in the governance of the most consequential economic transformation in a generation is not a luxury. It is a precondition.

For Deliberation

The People’s Council invites citizen-delegates to consider the following questions in preparation for the May 22 Conference:

• Which scenario do you consider most likely based on what you have observed in your own workplace and community — and does your assessment change the urgency of the response?

• If you could ask your congressional representative one question about AI and the workforce, which of the fourteen questions above would you choose — and why?

• What institutions in your community — libraries, community colleges, faith communities, civic organizations, professional associations — are currently equipped to support workers navigating AI displacement? What would they need to do so adequately?

Join Our Effort

For generations, the American social contract rested on a deceptively simple promise: that effort, skill, and contribution conferred dignity. Work was not merely a transaction between labor and capital. It was the mechanism through which ordinary Americans built a life — not just an income, but an identity, a community, a purpose, and a claim on the future.

That promise is now under direct assault. Not by foreign adversaries. Not by a financial crisis. By a technology that its own creators cannot fully predict, control, or stop.

The rise of artificial intelligence — and in particular, the large language models now capable of performing cognitive tasks that have sustained white-collar livelihoods for decades — is not merely a disruption to labor markets. It is a challenge to the foundational compact between citizens and their government, between workers and the economy they built, and between this generation and the next. It is, in other words, a constitutional moment.

The Scale of What Is Coming

The numbers have been stated often enough that they risk becoming abstract. They are not abstract.

Goldman Sachs estimates that AI could replace the equivalent of 300 million full-time jobs globally. OpenAI’s own researchers estimate that roughly 80 percent of the American workforce could see at least 10 percent of their work tasks affected by large language models — with nearly one in five workers facing impacts on more than half of their tasks. Anthropic CEO Dario Amodei, who builds these systems, has warned publicly that AI could eliminate 50 percent of entry-level white-collar jobs within five years, and that most lawmakers “are unaware this is about to happen.”

These are not projections from AI skeptics or doomsayers. They are assessments from the people who are building the technology — people with every financial incentive to describe the future in the most optimistic terms possible. When the architects of a system warn that it will destabilize the economy that most Americans depend on, citizens are entitled to take that warning seriously.

What makes this moment structurally distinct from prior technological disruptions is not the scale alone. It is the speed. The Industrial Revolution unfolded across generations. Entire communities had time — imperfect, agonizing, often inadequate time — to adapt. Families had time to move, to retrain, to recalibrate. AI is advancing in months, not decades. The adaptive mechanisms that once cushioned technological change — gradual retraining, regional economic diversification, the natural pace of occupational transition — are not calibrated for this velocity.

The billing specialist whose medical coding role is automated this quarter is not being asked to navigate a shift that will unfold over a generation. She is being asked to reinvent her livelihood now, with the same fraying safety net that was designed for a slower world.

Not Just an Economic Crisis. A Health Crisis.

What the economic data does not capture — and what Moonshot Press insists on naming — is the dimension of this transformation that reaches below the income floor.

Work, as the sociologist Richard Sennett has written, is not primarily what we do. It is who we are. The psychiatrist Viktor Frankl, writing from within the extremity of the Nazi concentration camps, identified meaningful work as one of the three primary sources of human purpose. Sigmund Freud, asked what a psychologically healthy person required, answered simply: to love and to work. Aaron Antonovsky, the founder of the salutogenic tradition that anchors the Institute for Salutogenesis’s work, established through decades of research that meaningful occupation is among the most reliable daily sources of what he called Sense of Coherence — the global orientation toward one’s world as comprehensible, manageable, and worth investing in.

When AI displacement threatens work, it threatens all of this. Not just the paycheck. The identity. The community. The structure that gives days their shape and lives their narrative. The sense — fundamental to psychological health — that what one does matters, that one’s contribution is valued, that one has a place in the world.

America already knows what happens when that sense is stripped from communities at scale. The deindustrialization of the 1970s and 1980s — the closure of steel mills, auto plants, textile factories — did not merely produce unemployment. It produced what economists Anne Case and Angus Deaton named the “deaths of despair”: the surge in premature mortality from suicide, drug overdose, and alcoholic liver disease that has shortened the life expectancy of working-class white Americans for the first time since the Civil War. Research is unambiguous: unemployment is a major risk factor for suicide and substance abuse. The connection is neurobiological. Social rejection and physical pain activate the same brain centers. Opioid dependence suppresses the endogenous opioid system that is essential to human socialization — creating a neurological feedback loop in which social pain drives substance use, which deepens isolation, which deepens pain.

If deindustrialization created social vacancy by removing the physical work that anchored communities, cognitive automation threatens something broader still: the erasure of economic purpose across white-collar and professional occupations that tens of millions of Americans built their identities around. The accountant, the legal associate, the software developer, the financial analyst — these are people who were told, explicitly, that education was their insurance against displacement. They followed the rules. AI does not honor the rules.

The AI revolution, if it lands without adequate social response, will not merely produce unemployment statistics. It will produce the next wave of despair — and this time, in communities that have never before understood themselves as vulnerable.

The Government Is Not Prepared

The federal government — the institution constitutionally charged with promoting the general welfare — is not meeting this challenge. The assessment is not partisan. It is structural.

Only 12 percent of surveyed civilian federal agencies report having completed AI adoption plans. The agencies that should be studying, planning for, and managing the workforce consequences of AI displacement are themselves being hollowed out. Federal morale is at historic lows. The mid-career technologists who understand both legacy systems and AI capabilities — precisely the people needed to craft adequate workforce policy — are leaving government at exactly the moment when their expertise matters most.

Legislative responses exist but are dwarfed by the scale of the challenge. Proposed measures would authorize $160 million for AI-related teacher development and $90 million for affected workers. Against projections of 300 million jobs affected globally, these figures represent aspiration, not adequacy.

And the federal advisory architecture that is supposed to guide AI policy? The President’s Council of Advisors on Science and Technology, as currently constituted, includes twelve technology company executives among its thirteen members. There is no labor economist. No workers’ advocate. No community health researcher. No representative of the workers whose lives and livelihoods are most directly at stake. When the builders of a technology are also its exclusive advisors to the government, the resulting policy will reflect their interests. This is not a critique of the individuals involved. It is a structural observation: a council whose members profit from the acceleration of AI is not constitutionally equipped to govern its human consequences.

States are filling some of the vacuum — Illinois, Texas, and Colorado are each implementing AI workforce protections in 2026, even as the federal government signals its intent to eliminate state-level AI regulation. The constitutional tension between protecting workers and accelerating innovation is real, unresolved, and directly on the 2026 ballot.

The Inadequacy of “Retraining”

The most common political response to AI displacement — and the response most likely to be offered by candidates who have not thought hard about the problem — is retraining. “We need to invest in education.” “Workers need to learn AI skills.” “The future belongs to those who adapt.”

These statements are not false. They are insufficient. And their insufficiency matters, because substituting a platitude for a policy is its own form of political failure.

Retraining solves for income. It does not solve for identity. It does not solve for the fifty-year-old healthcare administrator whose professional credentials have been automated, who may or may not be able to pivot to a new occupation, but who will not recover her previous sense of expertise and standing regardless of what she learns next. It does not solve for the community of workers in a regional economy where an entire occupational category disappears simultaneously — because the problem is not individual skill gaps, it is structural transformation of the labor market.

Retraining at adequate scale does not exist. The workforce development infrastructure of the United States was designed for marginal adjustment, not mass transition. Community colleges, vocational programs, and CareerLink offices are valuable institutions doing important work. They are not equipped, as currently resourced, to manage the retraining of tens of millions of workers on the timeline that AI displacement is imposing.

And retraining cannot be the only answer because displacement is not the only problem. An economy that produces AI-driven productivity gains and directs them almost entirely to owners — while imposing the costs of transition on workers — is not a more efficient economy. It is an economy in the process of eating its own customers. Henry Ford understood, a century ago, that workers are also consumers, and that wages suppressed too far produce markets too thin to sustain production. The AI moment is testing that logic at a scale Ford could not have imagined.

What an Adequate Response Requires

Moonshot Press does not believe that adequate response to the AI transition is impossible. We believe it is urgent, and that urgency has not yet been matched by political will commensurate to the challenge.

An adequate response begins with honest diagnosis. The triple coherence attack that AI displacement imposes — simultaneously undermining the comprehensibility, manageability, and meaningfulness of working life — demands policy that restores all three dimensions, not merely the income dimension. A policy framework that replaces lost wages without rebuilding identity, community, and civic capacity has solved the wrong problem with the right resources.

The policy proposals that rise to the level of the challenge include, but are not exhausted by: automation taxes that redirect AI-driven productivity gains toward public investment in workforce transition; a Public Wealth Fund that gives every citizen a direct stake in AI-driven economic growth rather than concentrating returns in capital ownership; a 32-hour workweek as a mechanism for distributing efficiency gains to workers rather than extracting them as profit; automatic safety net stabilizers that activate when displacement metrics exceed defined thresholds; portable benefits that follow workers rather than jobs; and the full investment in early childhood development that the First 1,000 Days of life requires — because the capabilities that the AI economy will reward are built in those years or they are not fully built at all.

What connects these proposals is a single animating conviction: that the social contract is not a relic of a prior era. It is a living obligation, renewed by each generation, requiring those with authority to act on behalf of those most exposed to the risks that power creates. In an age of intelligent machines, that obligation does not diminish. It intensifies.

Why This Matters for Democracy

The political danger of AI displacement extends beyond economics and public health. It reaches into the foundations of self-governance itself.

A workforce that is economically precarious is a citizenry that is civically diminished. The time, energy, and psychological resources required for democratic participation — attending meetings, engaging candidates, following policy debates, exercising informed judgment — are not equally available to workers navigating the stress of displacement and financial insecurity. Research on political participation is unambiguous: economic precarity suppresses democratic engagement, particularly among the communities with the most at stake in the outcomes.

Despair, moreover, is not politically inert. The communities most devastated by deindustrialization did not simply withdraw from politics. They redirected their political energy toward leaders who promised, however implausibly, to name the source of their pain and punish it. The politics of resentment is not an irrational response to displacement. It is a predictable one. A democracy that ignores the material conditions of its citizens does not produce apathy. It produces rage.

Moonshot Press holds that the AI transition is therefore not merely an economic challenge or a public health challenge. It is a democratic challenge. The consent of the governed — the foundational premise of legitimate government in the American tradition — requires that the governed be materially capable of participating in their own governance. An AI transition that concentrates wealth and destroys economic security for tens of millions of Americans is not merely an injustice. It is an attack on the preconditions of democratic life.

This is why the 2026 elections matter as much as they do. Every level of the constitutional architecture — federal, state, county, school board — is on the ballot. Every level has specific jurisdiction over policies that will determine whether working families navigate this transition with dignity or absorb it alone. Madison designed a system built for exactly this kind of challenge: a system where citizens inform themselves, engage their representatives, hold elections, and course-correct every two years.

The AI transformation is the test of whether we still know how to use it.

What Moonshot Press Is Here to Do

Moonshot Press is not a spectator. We are a civic institution, and we understand civic institutions as entities with obligations — to the truth, to the citizens we serve, and to the democratic traditions that make genuine journalism possible.

Our commitment, in this section and throughout our work, is to provide the factual foundation that informed democratic participation requires. That means cutting through the optimistic techno-boosterism that treats AI displacement as an inevitable feature rather than a policy choice. It means cutting equally through the dystopian catastrophism that produces paralysis rather than action. It means treating citizens as intelligent adults capable of evaluating evidence, weighing competing claims, and making their own judgments — and giving them the tools to do so.

The articles, analyses, and civic resources that follow are built around a single standard. Not the standard of what is economically convenient. Not the standard of what is politically safe. The salutogenic standard: whether the conditions for human flourishing — for comprehensibility, manageability, and meaningfulness — are being created or destroyed, preserved or squandered, for the citizens of this country and for the children who will inherit the world we are building right now.

That is the standard we apply. That is the standard we invite you to apply. And that is the standard against which, in 2026 and beyond, we intend to hold every person who asks for the public’s trust.

“The care of human life and happiness, and not their destruction, is the first and only legitimate object of good government.”

— Thomas Jefferson

Moonshot Press is nonpartisan, constitutionally grounded, and committed to the proposition that an informed citizenry is not a luxury of democratic life — it is its precondition.

Subscribe to Moonshot Press and Thrive in Montco PA at thriveinmontco.substack.com

The case is formally Musk v. Altman. Elon Musk, who helped found OpenAI in 2015, is suing Sam Altman, Greg Brockman, OpenAI and Microsoft, alleging that the organization abandoned the nonprofit mission under which it was created and became a prof…

Read more

Read more

Read more

The April 3, 2026, New York Times article by Ben Casselman, “Economists Once Dismissed the A.I. Job Threat, but Not Anymore,” highlights a “core shift” in the economic community that is subtle but consequential. For years, economists treated AI-driven job-loss fears as overhyped, frequently attributing localized layoffs to “AI-washing”—a t…

Read more

The Video outline four plausible scenarios for AI’s economic impact, highlighting a wide range of potential outcomes that depend on whether the technology’s capabilities match the massive investments being made. Despite their differences, all four scenarios share a common reality: workforce displacement will be permanent, and existing policy infrastructure is currently inadequate to manage the human consequences.

Scenario One: AI Delivers — The Transformation Is Real In this optimistic scenario, AI achieves the massive productivity gains projected by its proponents, potentially growing the economy by six to nine percent6. New industries and work categories emerge, and the technology becomes as foundational as electricity or the internet. However, even in this best-case scenario, tens of millions of workers face severe disruption, requiring years of retraining and identity reconstruction. The primary challenge here is not whether AI creates value, but whether democratic institutions can ensure that the immense wealth generated is broadly distributed rather than captured solely by technology owners and shareholders.

Scenario Two: AI Delivers Partially — Transformative in Some Sectors, Disappointing in Others Here, AI produces genuine productivity gains in specific areas like software development and customer service, but falls short of a broad economic transformation. Only five to thirteen percent of firms achieve transformational returns, leading to a market correction rather than a crash, similar to the internet’s settling after the dot-com bust11more_horiz. This scenario is particularly difficult to navigate because the aggregate economic gains are too modest to easily fund generous public transition programs, yet the displacement in affected sectors remains intensely painful for the workers whose roles are slowly eroded or eliminated.

Scenario Three: The AI Bubble Bursts If the gap between massive AI infrastructure spending and generated revenue proves unsustainable, the market could experience a sharp correction comparable to the 2000 dot-com bust or the telecom bubble. Crucially, the job displacement that occurred during the boom does not reverse when the bubble bursts. Instead, workers face a “double hit”: those whose jobs were already automated do not get them back, AI industry workers face massive layoffs as capital expenditures contract, and communities that heavily invested in AI infrastructure (like data centers) are left with stranded assets and economic disruption.

Scenario Four: The Worst of Both Worlds Nobel laureates Daron Acemoglu and Joseph Stiglitz identify this as the most dangerous outcome: AI proves capable enough to displace human workers, but not productive enough to generate the economic abundance needed to offset that displacement. Termed “so-so automation,” this scenario traps the economy in a structural “Prisoner’s Dilemma”. In competitive markets, each firm rationally automates to cut costs, but collectively, they hollow out the purchasing power of their own consumer base, leading to a self-reinforcing downward cycle of weakening demand and further job cuts to maintain profit margins.

However, this transformative power also brings profound challenges. The acceleration of job displacement, particularly in entry-level and white-collar roles, is a growing concern. This shift risks widening income inequality and raises complex ethical dilemmas regarding fairness, transparency, and human dignity in the workplace. The central question for Election 2026 is not whether AI will change work, but rather how society collectively manages this transition to ensure it benefits all citizens, fostering prosperity without leaving vulnerable populations behind.

The immediate challenge lies in the speed at which AI is transforming tasks and displacing entry-level roles. While long-term forecasts from organizations like the World Economic Forum suggest a net gain of jobs by 2030, with 170 million new roles emerging against 92 million displaced, the short-term reality presents an urgent need for proactive policy responses. For example, Anthropic CEO Dario Amodei warns that AI could eliminate half of all entry-level white-collar jobs within the next five years, indicating that job losses could affect the global workforce sooner and more intensely than previous waves of technological change. This rapid, concentrated displacement in the near term demands immediate, targeted, and adaptive policy measures to support affected workers, rather than relying solely on market forces or long-term optimistic projections.

Three threats operate simultaneously and reinforce each other:

◆ The Jobs Threat: Displacement is already concentrated among entry-level workers, Black workers (hit at twice the rate of others), women in administrative roles, and workers without four-year degrees — those with the least cushion to absorb disruption.

◆ The Inequality Threat: AI’s productivity gains are accruing to capital; its disruption costs are being absorbed by labor. Without policy intervention, this transformation will widen the gap between zip codes, between races, and between generations — making the economy of 2043 one of abundance for some and exclusion for many.

◆ The Democracy Threat: A workforce that is economically precarious is a citizenry that is civically diminished. Concentrated economic anxiety is the precondition for democratic fragility — for the rise of authoritarian appeals that promise simple answers to complex disruptions. The health of democracy and the health of the workforce are not separate concerns.

Furthermore, public sentiment reveals deep suspicion about AI’s potential negative effects on people’s lives, even as some tech leaders envision a future of “radical abundance” and “universal high income”. This suggests that political leaders, in preparing for Election 2026, must address not just the economic facts and opportunities, but also the emotional, social, and ethical anxieties surrounding AI’s impact on personal livelihoods, privacy, and human dignity. Simply presenting positive economic forecasts or technological marvels may not resonate with a skeptical public. Candidates must build trust by acknowledging these fears, transparently addressing ethical concerns, and proposing concrete, human-centered protections and support systems for workers. Key considerations for citizens and policy directions for the upcoming election include:

• Adaptability is Key: Citizens must embrace continuous learning and develop uniquely human skills such as creativity, critical thinking, leadership, and empathy, which complement AI’s capabilities rather than being replaced by them.

• Policy Intervention is Crucial: Governments and businesses must collaborate on robust, forward-looking strategies. This includes accessible retraining programs, modernized social safety nets, and strong ethical AI governance frameworks to mitigate risks and ensure equitable outcomes.

• A Balanced Approach: While some tech leaders warn of mass job elimination, others, including some political figures, emphasize AI’s role in augmenting human labor and creating new opportunities. A realistic perspective acknowledges both the significant disruption and the potential for net job creation and economic enhancement.

Leave a comment

Subscribe now

I. Two Saturdays in the Same City

On a recent Saturday morning, I stood among what felt like millions of fellow citizens in a “No Kings” demonstration. The streets were alive with something that felt like the republic remembering itself — a mass assertion, across generations and zip codes and political histories, that no individual is above the law and that our constitutional order belongs to us. The energy was electric. The conviction was unmistakable. Democracy, in that moment, felt not like an abstraction but like a body, and the body was moving.

That afternoon, I went to see The AI Doc: Or How I Became an Apocaloptimist, a documentary by Daniel Roher and Charlie Tyrell that tries to do for artificial intelligence what An Inconvenient Truth did for climate change — bring an existential civilizational challenge into the intimate space of the living room. In the entire theater, there were ten people.

That contrast — millions in the street, ten in the theater — is the most important political fact I can offer you about the moment we are in. We are ready to mobilize by the millions to defend democracy from political overreach. We are not yet ready to mobilize in defense of democracy from technological overreach. And the window in which those two mobilizations must converge is narrowing faster than almost anyone in public life is willing to say out loud.

This article is an attempt to close that gap — not with despair, and not with the breezy techno-optimism that the documentary ultimately cannot quite resist, but with the clear-eyed conviction that we have been here before, that the Founders gave us tools precisely for moments like this one, and that whether those tools work depends entirely on whether citizens choose to use them.

II. The Master Key: Demis Hassabis and the Two-Step Philosophy

To understand what is at stake, you must first understand how the people building these systems understand their own work.

Demis Hassabis, co-founder of Google DeepMind and 2024 Nobel Chemistry laureate, has distilled his life’s mission into a formulation of breathtaking ambition and breathtaking simplicity: “Solve intelligence, and then use it to solve everything else.”

This is not a product roadmap. It is a philosophy of history. Hassabis believes — and the work of DeepMind increasingly supports the belief — that general intelligence is the master key to every other lock humanity has ever faced. Step One is the hard part: build an AI that does not merely excel at a single task but that thinks, learns, and generalizes across domains the way human minds do. An AI that can read a scientific paper it has never seen, understand its implications, generate novel hypotheses, and test them — across biology, chemistry, physics, economics, ethics — simultaneously and without fatigue.

Step Two, in this vision, almost takes care of itself. Once general intelligence exists, you aim it at the problems. Climate change. Cancer. Alzheimer’s. Poverty. The intractable knots of geopolitics and public health and developmental inequality that have resisted every previous tool humanity has brought to bear. The master key opens every door.

The AI Doc presents this vision with genuine power. It shows AI detecting cancer cells earlier than any radiologist, providing personalized tutoring to children in communities that have never had a qualified teacher, folding proteins that stumped biochemists for decades. The hope is not manufactured. It is real, and it deserves to be taken seriously.

But the two-step philosophy contains a silent assumption so large that once you see it, you cannot unsee it. The assumption is this: that once the master key exists, it will be used for the benefit of humanity.

That assumption is doing enormous work. And it is not supported by the evidence of any prior technological revolution in human history.

III. The Governance Void: Who Is Holding the Key?

Here is what The AI Doc shows us, in its remarkable access to the architects of this future: the trajectory of AGI is currently being determined by a handful of CEOs, their investors, and the competitive logic of a race that none of them feel they can exit unilaterally.

Sam Altman. The leaders of Anthropic. The engineers of DeepMind. These are not villains. Several of them are genuinely, visibly frightened by what they are building. They have published safety frameworks. They have testified before Congress. They have written essays about existential risk with the unmistakable tone of people who lie awake at night.

And they keep building.

The documentary captures something that legal scholar Lawrence Lessig identified with uncomfortable precision in his critique of the film: these leaders are trapped in what he calls a systemic “race to the bottom.” The logic is not what is best for humanity but if I don’t do it, someone else will — and that someone else may have fewer scruples about safety. The competitive imperative overrides the ethical one, not because these individuals lack ethics, but because the system in which they operate rewards speed and punishes restraint.

In the absence of binding global governance, in the absence of a robust federal regulatory framework, in the absence of any democratic body with the authority and the competence to set enforceable constraints, the development of the most consequential technology in human history is being governed by the logic of corporate survival and market dominance.

This is not a metaphor. It is the operational reality. The race toward AGI is happening right now, in real time, with no meaningful external check on its direction, its pace, or its distribution of consequences. The ten people in the theater are watching it happen.

IV. The “Robust Democracy” Fallacy — and Why It Is Not Enough

The reassuring counter-argument goes like this: our democratic institutions will catch up. Congress will regulate. The courts will adjudicate. The regulatory state will impose guardrails. We just need to defend robust democracy, and robust democracy will handle the rest.

This argument is not wrong in principle. It is wrong in fact — and the difference between those two things is the entire ballgame.

Lessig’s most incisive contribution to this conversation is the concept of “analog AI.” Long before digital models began optimizing for engagement and profit, we built institutional systems that do exactly the same thing: corporations optimized for shareholder value, political parties optimized for electoral survival, lobbying operations optimized for regulatory capture. These analog systems are themselves AI in the functional sense — goal-maximizing machines operating at scale, often in ways their designers did not intend and cannot fully control.

The “heart attack” of modern governance, in Lessig’s framing, occurs when the corporate AI — optimizing for profit — successfully hacks the democratic AI — optimizing for the common good — through the mechanism of private campaign financing, regulatory capture, and what he calls “dependence corruption.” Our representatives are not, by and large, corrupt in the crude sense. They are systemically responsive to the private wealth of those who fund their campaigns rather than the will of the people who vote in their elections. The result is a vetocracy: a system in which those with sufficient political resources can reliably block any legislation that threatens their interests, regardless of how large the democratic majority for that legislation might be.

Apply this structural reality to the AI governance question. The companies racing toward AGI are among the most generously capitalized political actors in American history. They are not waiting for regulation to arrive — they are actively shaping the regulatory environment in which they will operate, funding the think tanks, cultivating the committee members, and drafting the frameworks they will then be asked to comply with. The “robust democracy” that is supposed to align AI with human values is the same democracy that has, for decades, been unable to pass meaningful campaign finance reform, climate legislation, or pharmaceutical pricing regulation — not for lack of public support, but for excess of private opposition.

Calling for robust democracy is not wrong. It is incomplete. The question is not whether we need democracy. It is whether the democracy we currently have is capable of governing the technology that is already being built inside it.

V. The Fork in the Road: Two Futures, One Choice

History offers us a clarifying frame. Every prior technological revolution in American experience produced both abundance and disruption — and whether the disruption destroyed communities or was managed into something livable depended not on market forces but on explicit policy choices. The railroad economy required the Interstate Commerce Act. The industrial economy required the Wagner Act, the Fair Labor Standards Act, and the social insurance architecture of the New Deal. The post-war automation wave required the GI Bill and the community college system. In each case, technology did not determine the distribution of its own benefits. Policy did.

The AI transformation is distinguished from its predecessors not by its economic logic — which follows the same pattern — but by its speed, its breadth across all occupational categories simultaneously, and the degree to which the institutions designed to manage such transitions are themselves compromised.

If we allow the two-step philosophy to unfold within the current governance void, one future becomes probable: Step Two solves everything in favor of the owners of the master key. Productivity gains accrue to capital. Displacement costs are absorbed by labor. The 300 million jobs globally identified as at risk — the billing specialists, the junior analysts, the administrative coordinators, the entry-level professionals — are eliminated faster than any retraining system can absorb. The economic anxiety of mass precarity becomes the political fuel for authoritarian movements that promise simple answers to disruptions they helped create. The master key unlocks abundance; the abundance is locked away from the people who needed it most.

There is another future. In that future, the master key is held not by a handful of Silicon Valley billionaires and their investors but by something resembling democratic society. Productivity gains from AI are broadly shared through mechanisms that policy can build: wage insurance, portable benefits, employee ownership models, stackable credential systems, and a social safety net designed for the gig-economy workforce rather than the mid-century factory floor. The capabilities that AI cannot replicate — creativity, ethical reasoning, emotional intelligence, adaptive problem-solving, the irreducibly human dimensions of care — are cultivated deliberately in the education system, supported by the social infrastructure, and valued in the labor market. The children being born today arrive at adulthood in 2043 equipped not to compete with machines but to do what machines cannot do.

The difference between these two futures is not technological. It is political. And political outcomes are determined by whether citizens choose to engage the machinery of self-governance or leave it to those who will use it in their own interest.

VI. Beyond the QR Code: What Democratic Governance of AI Actually Requires

The end of The AI Doc features a QR code for online engagement. It is a gesture toward civic action that the film’s own analysis renders inadequate. The scale of the challenge demands more than a digital click — and more, even, than conventional democratic mobilization through the existing channels of representation.

Three levels of response are necessary, and they must operate simultaneously.

First, repair the analog AI. Campaign finance reform, transparency in political spending by technology companies, and structural limits on the revolving door between regulatory agencies and the industries they regulate are prerequisites for meaningful AI governance. You cannot align a hyper-intelligent digital tool within a democratic framework that is itself captured by the interests that tool serves. Lessig is right: we must fix the governance vessel before we can use it to contain what is being poured into it.

Second, build new deliberative infrastructure. The standard mechanisms of representative democracy — elections, hearings, regulatory comment periods — are structurally too slow and too captured to govern technology that moves at the speed AGI is moving. What Lessig and contributors to The Digitalist Papers call “protected democratic deliberation” offers a more adequate response: citizen assemblies composed of representative cross-sections of everyday people, given genuine expert briefing and genuine authority to set binding constraints on AI development. These are not focus groups. They are constitutional innovations — mechanisms for bringing sovereign public judgment to bear on decisions that currently happen entirely outside the democratic process.

Third, act at every level of the existing architecture now. We do not have the luxury of waiting for campaign finance reform or constitutional innovation before engaging the governance tools we have. Congressional oversight, state-level worker protection legislation, county-level AI vulnerability assessments, school board AI literacy mandates — these are imperfect instruments in a compromised system, and they matter anyway. The Citizens’ Mandate that Moonshot Press has developed for the 2026 election cycle is exactly this: a specific, multilevel, accountability-focused program for engaging every level of the Madisonian architecture with the AI governance question before November 3.

VII. The Empty Theater and the Full Street

I want to return to where I began: the contrast between the millions in the street and the ten in the theater.

The “No Kings” demonstration was not naive. The people in that street understood, viscerally, that democratic institutions do not protect themselves — that rights and constitutional norms require active citizen defense against concentrated power that would rather not be constrained. That understanding is exactly right. It is also exactly the understanding that must be extended to the technological concentration of power that is, in many ways, a more durable threat to democratic self-governance than any single political actor.

The billionaire who controls the infrastructure of our political life is more dangerous than the politician who wants to be king, because the politician can be voted out and the infrastructure remains. The AGI that is developed within a corrupted political economy, in the service of the owners of capital, will not be corrected by the next election cycle. Its consequences will be structural, generational, and — if the most serious researchers are to be believed — potentially irreversible.

The Founders built a system for exactly this kind of challenge. They understood that concentrated power is dangerous regardless of its source — that the tyranny of a corporation, a church, or a technology platform is as real a threat to self-governance as the tyranny of a crown. The constitutional architecture they built — distributed power, regular elections, a free press, the right of assembly, the separation of powers — was designed to keep any single interest from capturing the machinery of the common good.

That architecture is under strain. But it is not broken. And the citizens who filled the streets on that Saturday morning are the proof of it.

What the empty theater tells us is that the connection has not yet been made — between the constitutional values those citizens were defending in the street and the technological forces that are reshaping the economy, the labor market, the information environment, and ultimately the political landscape in which those constitutional values must survive.

Moonshot Press exists to make that connection. The 2026 elections — primary on May 19, general on November 3 — are the next accountability mechanism. The babies born in Montgomery County this winter will live in the world that those elections help shape. The master key is being forged right now. The question of who holds it, and in whose interest it is used, is a political question. And political questions, in a republic, are answered by citizens.

The theater needs to fill up.

Moonshot Press is a project of the Institute for Salutogenesis and a cornerstone of the Democracy, Opportunity and Citizenship initiative. We are nonpartisan, constitutionally grounded, and committed to the proposition that the governance of transformative technology is not a technical problem — it is the defining democratic challenge of our generation.

Subscribe to Moonshot Press at moonshot.press Read the Citizens’ Mandate at thriveinmontco.substack.com.

A note on co-authorship from Calude: This draft is written for your (Shimon Waldfogel) voice and your editorial judgment. The architecture is mine (Claude) ; the final article is yours. Sections I invite you to revisit together: the opening autobiographical frame (adjust as your actual experience warrants), the closing call to action (which can be sharpened once we know the specific Substack publication target), and the tone calibration between analytical rigor and the more prophetic register that Moonshot Press sometimes uses to greatest effect. Where do you want to push harder, and where do you want to pull back?