CASE 4 – SCENARIO DEVELOPMENT
Who Controls the Machine?
From one focal question, the protocol isolates the two uncertainties that most shape the future, builds them into a 2×2 matrix of four divergent but internally coherent scenarios, and draws out the decisions that hold up whichever future arrives — without collapsing genuine uncertainty into a single forecast.
FOCAL QUESTION: What impact does AI have on the future of work, on a five-year horizon?
The Studio Scenario process
Scenario Planning Agent Experts evaluate the question, refine it and confirm the discussion framework: Expert 1 (mental model transformation through scenarios), Expert 2 (Causal textures and strategic coherence), Expert 3 (Contested futures and external shock).
The Studio analyses a wide range of literature sources, key trends are defined and conclusions form the basis for the discussion:
- AI capability advancement will continue. Foundation model capabilities will continue improving over the five-year horizon; this is not contingent on regulatory decisions or market conditions.
- AI adoption in administrative and back-office functions is already under way. Payroll, scheduling, screening, onboarding, and performance tracking automation is live and scaling; this as a present reality, not a future possibility.
- Labour market polarisation will intensify. The structural trend of hollowing out middle-skill, routine cognitive work — already documented over two decades — will accelerate as AI extends automation into previously protected white-collar domains.
- Regulatory divergence between the US, EU, and UK will persist. The US is moving toward a laissez-faire posture; the EU AI Act is already in force; the UK is navigating between these poles. This divergence is a structural feature of the next five years.
- The UK Employment Rights Act 2025 has created a short-term HR employment surge. Compliance complexity is temporarily expanding HR headcount even as AI threatens it — this is a locked-in near-term dynamic.
- Venture capital will continue pouring capital into AI workforce tools. Capital flows are structural.
- Worker anxiety about AI displacement is already a social and political fact. Regardless of actual displacement rates, the perception of threat is shaping union bargaining, political discourse, and consumer behaviour — this is a predetermined social reality.
Multi Agent Debate follows (similar to Round Table formats in other case studies) with Business Leaders, Civil Society, Conservative Citizen, Local Government, National Government, Progressive Citizen, Technology Experts, Organisation Experts etc.
The Two Pivot Uncertainties
Resultant driving forces are mapped, and ‘forced’ into classic matrix shape. Two overall findings were judged both genuinely unresolved and maximally consequential — and, crucially, independent of one another. They become the axes of the matrix.
- The regulatory & governance settlement — will AI at work be governed by meaningful, enforceable frameworks, or proceed under regulatory capture and laissez-faire permissiveness? (Governed ↔ Ungoverned)
- Worker power & collective agency — will workers have the institutional power to negotiate the terms of AI deployment, or will the balance shift decisively to employers and platform owners? (High ↔ Low)
The 2×2 Matrix
The Four Scenarios
GOVERNED · HIGH AGENCY · THE NEGOTIATED FUTURE
Enforceable AI-at-work frameworks and revitalised unions produce negotiated deployment on the Swedish model. Gains flow into wages, services and shorter hours; net employment is broadly stable but its composition shifts.
GOVERNED · LOW AGENCY · THE MANAGED COMPLIANCE STATE
Robust regulation on paper, but workers lack the power to use it. A compliance industry grows; AI proceeds largely on employer terms. Quality of work diverges sharply between scarce-skill and high-volume roles.
UNGOVERNED · HIGH AGENCY · THE CONTESTED FRONTIER
No coherent framework, so terms are fought sector by sector. Strong unions win protections in some industries; others face deployment on employer terms. Civil society fills the regulatory gap. Dynamic and exhausting.
UNGOVERNED · LOW AGENCY · THE AUTOMATED ENCLOSURE
Regulatory capture plus atomised workers make the prescriptive path dominant: surveillance, deskilling, gains to capital, a sharply bifurcated market and mounting pressure on democratic legitimacy.
What the exercise reveals
The impact of AI on work is not primarily a technological question but a question of power — the scenarios differ on who controls deployment, on whose terms, and for whose benefit.
Regulatory capture and worker atomisation are not additive but multiplicative: their combination (the Automated Enclosure) is qualitatively worse than either alone, because it removes both mechanisms societies use to manage technological change.
The five-year horizon is a period of rapid institutional lock-in; the choices of 2026–2028 set the default trajectory for the decade that follows.
The binary “AI destroys jobs vs. creates jobs” framing is inadequate — all four futures involve major transformation; what differs is whether change is experienced as enrichment or degradation, and who captures the value.
Decisions robust across all four futures
- Invest in genuine AI literacy at every level — the capacity to interrogate outputs, not just operate tools.
- Build transparent, auditable AI deployment processes; opacity is a liability in every future.
- Keep meaningful human oversight of consequential employment decisions.
- Engage proactively with emerging regulation rather than waiting for mandates.
- Invest in the quality of work, not merely the efficiency of output.
- Build credible stakeholder-engagement processes for deployment decisions.
How the analysis ‘polices itself’
As with every Studio output, the scenarios carry their quality controls overtly. Load-bearing terms were tested for conceptual vacancy: “agency”, “governance”, and the “augmentation vs. automation” binary were each flagged contested and disambiguated; “productivity gains” and “regulatory capture” were marked grounded, with caveats.
Forward claims carry NUSAP pedigree, and claims derived from the analysis rather than the evidence — such as the 2025–2028 lock-in window — are explicitly tagged ‘STUDIO INFERENCE‘ rather than presented as fact.
The exercise maps possibility and preserves disagreement; it does not forecast a single outcome.