The original New Deal social contract was built on an assumption: that the American economy would produce enough jobs, paying enough wages, to sustain a working class that could afford housing, raise children, and retire with dignity. The institutional apparatus built around that assumption — Social Security, employer-provided healthcare, the forty-hour week, the minimum wage — made sense when the assumption held.
The assumption is no longer holding. And the institutional apparatus designed around it is becoming, in some respects, an obstacle to the adaptations that would serve people better.
What the Old Contract Actually Said
The postwar social contract was never written down. It was an implicit arrangement: you participate in the labor market, you receive wages, and those wages plus a set of social programs provide the floor below which your life will not fall. Unemployment is temporary; retraining leads to re-employment; the system is designed for disruption but not for permanent displacement.
This arrangement had significant exclusions from the beginning — it was built around the male industrial worker, and it left out domestic workers, agricultural laborers, and most women and people of color in its early decades. But its logic was expansible, and it did expand, slowly and imperfectly, to include more people over the postwar period.
The new disruption is different in kind from what the contract was designed to absorb. It is not cyclical unemployment — the temporary displacement of workers who will be rehired when demand recovers. It is structural displacement — the permanent reduction in demand for specific forms of human labor by technology that performs the same functions at lower cost. Social insurance systems designed for cyclical disruption are poorly equipped for structural disruption.
What Security Needs to Look Like Now
A functional social contract for the AI age needs to accomplish three things that the current one does not.
First, it needs to decouple basic security from employment status. Healthcare in the United States remains tied to employment in ways that make job loss acutely catastrophic and suppress the labor mobility that productive economies require. In a world where AI-driven displacement is structural rather than cyclical, this link needs to be severed. Universal healthcare — not as an ideological commitment but as an engineering requirement for a functional labor market — is a prerequisite for managing the transition well.
Second, it needs to distribute the productivity gains from AI broadly rather than concentrating them. AI systems are generating enormous productivity gains. Those gains are currently flowing almost entirely to the owners of the systems — the technology companies, their employees, and their investors. A social contract that allows the majority of AI's productivity to accrue to a small minority while the majority of workers bear the costs of disruption is not stable.
The mechanisms for broader distribution are still being designed: sovereign AI wealth funds, data dividends, expanded public ownership of AI infrastructure, strengthened taxation of AI-generated returns. None of these is a complete solution. All of them represent more serious engagement with the distribution problem than the current conversation offers.
Third, it needs to provide meaningful support for the period of transition — which is not weeks or months but potentially years or decades. Current unemployment systems in most developed countries are designed for gaps of three to six months. The retraining timelines for workers displaced from mid-career professional roles are typically two to five years, assuming retraining is even viable. The mismatch is not a detail.
The Political Obstacles
The new social contract needed for the AI age faces political obstacles that are not primarily ideological. They are structural.
The people who would most benefit from a redesigned contract — mid-career workers facing displacement, people in precarious employment, young workers entering a disrupted labor market — are not the people who set political agendas. The people who would face significant costs from redesigning the contract — the owners of AI infrastructure, the financial interests benefiting from current arrangements, the insurance and healthcare industries embedded in the employer-based system — are very well-organized and very politically active.
This is not a conspiracy. It's an interest alignment problem. The groups whose interests favor the status quo are better organized, better funded, and have clearer advocacy goals than the groups whose interests favor change. This is the normal condition of politics. Structural change happens anyway, occasionally, when disruption becomes too visible to manage and political conditions align.
The AI transition may eventually produce those conditions. The question is whether it does so soon enough to shape the institutional arrangements while they're still plastic, or late enough that the path dependencies have set in ways that make fundamental redesign prohibitively expensive.
What a New Contract Might Include
Describing a fully designed new social contract is beyond any single book or essay. But the outlines of what a serious response might include are becoming clearer through the policy conversations happening in Europe, in progressive economics, and in the emerging discourse around AI governance.
A serious response would include: portability of benefits across employment relationships; public healthcare not dependent on employer status; income support for workers in extended transitions, funded in part by the productivity gains AI generates; investment in the caring economy — education, healthcare, elder care — where human labor remains both irreplaceable and undervalued; and new forms of public ownership or oversight of AI infrastructure that distribute its returns more broadly.
None of this is radical relative to the scale of the disruption. All of it is politically difficult relative to the current institutional arrangements.
The old bargain is breaking. What replaces it is still being decided — not in a single policy debate, but in the accumulation of decisions, defaults, and political outcomes happening right now. After Work is an attempt to name what's at stake and to make the case that the decision matters, and that the window for making it deliberately is narrower than it looks.