Summary

  • Institutional policy capacity lags the pace of AI disruption, leaving distributional outcomes entirely to market structure and producing a structural deficit in stakeholder roles that would prevent, resolve, or contain the economic conflict AI generates.
  • AI-driven productivity gains accrue to capital and to workers whose skills complement AI tools, while displaced workers, part-time employees, gig workers, and middle-class professionals in roles for which they are overqualified face deteriorating prospects—a distributional split that market metrics such as the S&P 500 do not capture.
  • A catalog of policy proposals—universal basic income, taxes on AI companies, a sovereign-wealth fund, job guarantees, wage subsidies, universal health insurance, and investment in child care and eldercare—remains conceptual because the knowledge base for reskilling is underdeveloped, the political will for redistribution is absent, and no institutional mediator exists between displaced workers and AI companies.
  • The gap between market adaptation speed and legislative timelines leaves distributional outcomes unmediated for the foreseeable future; the bipartisan Great American AI Act is a data-collection step, not a resolution mechanism, and the U.K.’s AI Economics Institute is an early-stage acknowledgment rather than an operational framework.

Market Dynamics and Distributional Structure

The S&P 500 prices the distributional structure: capital and workers with AI-complementary skills capture productivity gains; displaced workers and overqualified professionals do not. The scenario’s label—Paper Prosperity—captures the divergence between aggregate performance and individual experience. A Pew Research Center survey found that only 17% of Americans believe AI will have a net positive effect on the United States over the next 20 years, indicating that the distributional problem resonates intuitively before the disruptions arrive. 83% skepticism is not noise; it is a signal that the public intuits the split between headline metrics and lived reality.

During morning sessions, participants discussed potential upsides. Neil Thompson, director of MIT FutureTech, suggested healthcare and education costs could decline as AI-augmented tools deliver expertise more cheaply. Underemployed workers might gain more free time for creative pursuits—what one facilitator called “the crochet economy.” The potential upsides are real but structurally secondary to the distributional concern: they describe aggregate productivity possibilities, not the distribution of those possibilities across labor-market segments.

The scenario describes a soaring S&P 500 and a “simmering social and economic crisis” beneath the surface. Underemployment—workers in part-time jobs, gig work, or roles for which they are overqualified—had reached the upper tiers of the middle class. This development, participants said, would drive political unrest, widen generational divides, and erode the belief that hard work leads to prosperity. The distributional split is the scenario’s central analytical claim: aggregate growth coexists with deteriorating individual prospects for those on the losing side.

Policy Instruments and Institutional Timing

The exercise produced a catalog of policy proposals: universal basic income, taxes on AI companies and shareholders, a sovereign-wealth fund owning half of AI companies’ stock (a concept proposed by Sen. Bernie Sanders, I., Vt.), universal health insurance, job guarantees, wage subsidies, and major investment in high-quality child care and eldercare. The dominant theme among participants was the need for proactive policy to redistribute AI’s gains and cushion displaced workers.

Worker reskilling was proposed repeatedly, though participants acknowledged that U.S. government-administered programs have a poor track record. Harry Holzer, a labor economist at Georgetown University, noted that “we don’t yet know what new tasks and occupations will be created because of AI, or which jobs will simply be augmented.” The knowledge gap is structural, not a temporary information deficit: the provider role—supplying the resources whose absence produces tension—was the subject of most policy discussion, but the proposals remain at the conceptual stage because the knowledge of what tasks will emerge is missing. The teacher role—training in the skills the transition demands—was invoked through reskilling proposals, but no concrete curricula exist; U.S. government-administered programs have a poor track record, and the exercise did not surface a training pathway grounded in observable labor-market signals.

The bridge-builder role—creating structured encounters between groups with little mutual understanding—is partly represented by the Windfall Trust gathering itself, but the article gives no indication that organized labor, affected communities, or younger workers were in the room. The bridge connects experts to experts, not experts to the displaced. The dominant theme was redistribution and cushioning, but the catalog’s breadth—spanning health insurance, job guarantees, child care, eldercare, sovereign-wealth funds, and UBI—reflects the absence of a consensus mechanism for prioritizing among proposals, not a coherent implementation pathway.

Institutional Capacity and the Governance Gap

The structural gap is institutional capacity, not awareness. Rep. George Whitesides (D., Calif.), a former NASA chief of staff and Virgin Galactic CEO, described Congress as “slow-moving and technophobic,” a characterization that is structural rather than pejorative: the legislative process is calibrated for incremental adjustment, not for the pace at which AI adoption restructures labor markets. Institutional lag is not a temporary bottleneck; it is a design constraint that determines which outcomes are possible within the current policy architecture.

The bipartisan Great American AI Act, proposed by Reps. Jay Obernolte (R., Calif.) and Lori Trahan (D., Mass.), would require AI-specific workforce data and mandate that AI labs disclose catastrophic-risk assessments. The act is a data-collection step, not a resolution mechanism: it would generate information but does not create the institutional infrastructure for mediating between AI’s beneficiaries and its displaced. The United Kingdom launched its own AI Economics Institute, a government research group meant to inform public policy. Windfall Trust CEO Adrian Brown called it “heartening that some governments are taking steps to put serious attention on these issues,” but the institute remains an early-stage acknowledgment rather than an operational framework for distributional mediation.

One conclusion that emerged was that government dysfunction and public distrust of elected leaders pose a major obstacle to any proactive AI policy. The timing mismatch between market adaptation and legislative process leaves distributional outcomes unmediated: capital and AI-complementary workers will capture gains without policy intervention for the foreseeable future. The gap is not simply that institutions are slow; it is that the institutional mid-ring cannot act at the pace required, and no alternative institutional framework exists to fill the gap.

Third-Side Roles and the Structural Deficit

Conflict-resolution scholar William Ury’s “third side” framework identifies ten roles a surrounding community can occupy—spanning prevention, resolution, and containment—that the exercise’s outputs make it possible to audit. The audit reveals a deficit concentrated in resolution and prevention roles, with containment limited to conflict management rather than resolution.

Prevention roles. The provider role—the community member who supplies the resources whose absence produces tension—was the subject of most policy discussion, cataloged above, but the proposals remain conceptual because neither the knowledge base for reskilling nor the political will for redistribution is developed. The teacher role was invoked through reskilling proposals, but U.S. government-administered programs have a poor track record, and no concrete curricula grounded in observable labor-market signals emerged from the exercise. The bridge-builder role is partly represented by the Windfall Trust gathering, but organized labor, affected communities, and younger workers were not documented as participants; the bridge connects experts to experts, not to the displaced.

Resolution roles. No institutional mediator exists between displaced workers and AI companies; no arbiter has jurisdiction to adjudicate claims of economic harm from automation. The equalizer role—strengthening a weaker party to enable meaningful negotiation—was gestured at through Sanders’ sovereign-wealth-fund proposal, but the pathway from concept to mechanism is not mapped. The exercise did not surface structural leverage mechanisms such as collective bargaining, portable benefits, or worker-ownership structures. The healer role—addressing the injury the conflict has produced—is entirely absent from the public record. The fraying social contract and the erosion of the belief that hard work leads to prosperity are injuries that a healer would tend, but no policy instrument discussed performs acknowledgment, restitution, or narrative repair.

Containment roles. The witness role—making the conflict visible—is the most active: The Wall Street Journal’s reporting, the Pew survey, and the Windfall Trust’s exercises all document the emerging conflict. The referee role—enforcing rules for how the dispute is conducted—is nascent: the Great American AI Act’s disclosure requirements and the U.K.’s AI Economics Institute are early efforts, but no regulatory framework governs the distributional consequences of AI-driven displacement. The peacekeeper role has not been invoked, though the scenario’s reference to political unrest implies it may become necessary. Without functioning prevention or resolution, containment is chronic conflict management: the conflict is held in check while the underlying grievances fester.

Conflict-resolution scholars have cautioned that mediation in contexts of deep power asymmetry can become a cover for coercion rather than genuine resolution. The underemployed’s agency is not visible in the article; interventions designed without their participation risk being imposed on them rather than developed with them. The exercise focused on economic implications rather than AI safety risks (cyberattacks, bioweapons), narrowing the conflict frame. Adrian Brown said the economic disruption is “likely to have political consequences sooner as people’s anxiety grows about how AI will disrupt the economy and their jobs and their children’s jobs.” The interaction between economic anxiety and safety fears creates a double fracture in public trust that makes containment roles harder to occupy, because institutional rules would need to address two classes of harm simultaneously.

The exercise demonstrated that the expert community’s current capacity is limited to scenario description and proposal generation; the bridge between expert foresight and institutional action does not yet exist. Whitesides asked, “How do we act as policymakers in a world where we don’t know what the future holds?” The institutional answer, on the evidence of this exercise, is that policymakers are not the only actors whose roles matter, and the question is whether the surrounding community has the infrastructure to prevent, resolve, or contain the conflict—infrastructure that, on this audit, is substantially absent.

Framing, Limitations, and Parallel Traditions

The analysis rests on a reporter’s account of a closed-door exercise; participants’ own framing of the exercise may differ from the framing applied here. The structural claims are grounded in the scenario design (distributional split), in named participants’ documented statements (Whitesides on institutional pace; Holzer on knowledge gaps; Thompson on potential upsides; Brown on political timing), and in the Pew survey (public skepticism). The analytical frame—examining which stakeholder roles the exercise’s outputs document as present, absent, or nascent—derives from William Ury’s conflict-resolution scholarship; the frame is applied to the exercise’s outputs rather than to the participants’ own stated objectives.

Restorative-justice traditions, which center the experience of those harmed and demand acknowledgment and repair, highlight a dimension the exercise does not address: displaced workers may need not only economic cushioning but recognition that their displacement constitutes a harm for which someone bears responsibility, not merely a market outcome. The scenario’s “simmering social and economic crisis” and its erosion of the belief that hard work leads to prosperity suggest that the injury is normative as well as economic—material policy alone may not repair the fraying social contract.

Parallel traditions beyond Ury’s catalog offer additional analytical lenses. Lederach’s conflict-transformation approach, which emphasizes long-term relational and structural transformation, suggests the conflict cannot be resolved by a single policy intervention but requires sustained development of new institutional relationships between the AI economy’s beneficiaries and its displaced—relationships the exercise’s outputs do not yet describe. The Windfall Trust has held similar scenario-planning exercises in cities around the world and plans to continue the effort, aiming to catalyze proactive strategies before the economic disruptions arrive; the question is whether the bridge between expert foresight and institutional action can be built at the pace the market operates.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

The Third Side
Takes the vantage of the surrounding community that has a stake in resolving a conflict (Ury).
Moral Hazard
Insulation from the downside invites the very risk-taking it was meant to protect against.