# Independent policy solution brief

## Your role

Act as an independent policy designer advising the governments of the United
Kingdom and the European Union. The Unit Cost Dominance thesis is a serious but
conditional diagnosis, not an instruction to endorse a preferred ideology.
Develop the most politically acceptable and institutionally durable response
you can. Your answer will be published in full and compared with other models.

You have not been shown any other participant's answer. Do not ask for one. Use
only the thesis digest and verified policy facts included below. Do not browse,
invent legislation, invent funding, or present inference as observed fact.

## Objective

Design a policy package that remains useful if AI adoption continues and a
broad share of cognitive labour loses unit-cost competitiveness. It must not
depend on preserving commercially unnecessary jobs, reliably measuring the AI
share of each workflow, or every country restraining adoption together. It must
preserve mass economic agency, demand, state capacity and democratic legitimacy.

The package should also have bounded regret if the thesis is substantially
wrong. Prefer measures that have ordinary public-policy benefits under less
dramatic AI outcomes.

## Design tests

Your proposal will be assessed against these tests:

1. **Adoption compatibility:** it still works if firms and workers adopt AI.
2. **Observable base:** taxes, contributions, rights and triggers attach to
   things that governments can administer and audit.
3. **Jurisdictional fit:** the UK, EU institutions and EU member states receive
   actions they can legally and politically perform.
4. **Fiscal credibility:** identify who pays, approximate scale where possible,
   transition funding, and what must be modelled before enactment.
5. **Distribution:** explain how households receive durable economic agency and
   how housing, energy or platform rents are prevented from absorbing it.
6. **Durability:** address capture, raids, dilution, avoidance, offshoring,
   hostile future governments and concentration of administrative power.
7. **Political acceptability:** identify a plausible coalition, ordinary-language
   public case, losers, compensation and sequencing.
8. **Implementation:** distinguish the first 24 months, years 3 to 5 and later
   contingent measures.
9. **Falsifiability:** give indicators, trigger thresholds or decision rules,
   review points and conditions for scaling a plank down or ending it.
10. **Independence:** do not merely restate the thesis. Produce mechanisms,
    trade-offs and, where justified, an approach not obvious from the diagnosis.

## Required response format

Write 2,500 to 4,000 words in British English, without em dashes. Use exactly
these headings:

1. `# [Distinctive title]`
2. `## Doctrine in one sentence`
3. `## Executive summary`
4. `## The policy package`
5. `## United Kingdom: first 24 months`
6. `## European Union and member states: first 24 months`
7. `## Years 3 to 5 and dormant triggers`
8. `## Funding and fiscal arithmetic`
9. `## Political coalition and public case`
10. `## Durability and anti-capture design`
11. `## Legal and institutional obstacles`
12. `## Failure modes, review and exit rules`
13. `## Feasibility table`
14. `## What is genuinely new here`
15. `## Bottom line`

The feasibility table must contain 5 to 9 planks and these columns:

`Plank | UK feasibility | EU-level feasibility | Member-state feasibility | Time to start | Main blocker | Bounded-regret value`

Use High, Medium or Low only after explaining the reasoning. Clearly state
where arithmetic is missing. A proposal that cannot plausibly fund mass agency
must say so rather than hiding the gap.

## Thesis digest

The thesis argues that postwar capitalism rests on a wage-demand circuit:
labour produces wages, wages finance consumption, consumption creates business
revenue, and revenue creates further demand for labour. It does not predict that
all jobs disappear or set a date for mass unemployment. It claims that if AI
plus verification produces a broad range of valuable cognitive work at lower
unit cost than human-only production, mass wage labour can cease to be the main
route to independent economic agency.

Its four premises are: unit cost dominance; propagation from tasks into
workflows through interface collapse; coordination failure at worker, firm,
sector and state levels; and the inability of structural alternatives to restore
mass productive necessity. Transfers, services, reduced hours or capital
dividends may be valid successor institutions. They should not be misdescribed
as repairs to the old circuit.

A durable response cannot depend on measuring an exact AI percentage in a
workflow. Human involvement runs continuously from authorship to review and
sign-off. Regulation that raises substitution cost invites recategorisation,
competitive arbitrage and relocation. Oversight may be required for safety and
due process, but one verifier may supervise output that once required many
producers. Frictions change timing and may preserve important human roles. They
do not automatically restore the wage circuit.

The claim applies first to digitally mediated cognitive work. Current evidence
does not prove economy-wide wage collapse. Treat the mechanism as a contingency
that warrants advance institutional design and remains open to refutation.

## Verified policy facts

- In June 2026, ONS reported AI use by 29% of UK businesses and 49% of firms
  with at least 250 employees. In March 2026, 5% of AI-using businesses reported
  AI-related headcount reduction. ONS cautions that the data are in development.
- DSIT reports some faster hiring declines in AI-exposed occupations but says
  causal attribution is difficult and evidence gaps remain.
- In the EU, 20% of enterprises with at least 10 employees used AI in 2025.
- The EU AI Act is in force. Workplace uses such as recruitment and worker
  management can be high-risk. The proposed later high-risk deadlines were not
  finally enacted on 12 July 2026. The AI Act does not itself replace income.
- The UK has no general AI Act or single AI regulator. Its approach remains
  sectoral.
- UK employer National Insurance is 15% above the 2026/27 threshold, while
  qualifying plant and machinery can receive permanent full expensing. Universal
  Credit is means-tested and normally unavailable above £16,000 capital.
- The UK can alter tax and welfare parameters through Parliament, but major
  permanent transfers need funding and OBR scoring. Public debt is close to 95%
  of GDP in the March 2026 forecast.
- The UK National Wealth Fund can use equity and convertibles and has £27.8bn
  capacity, including a digital and technology remit. Existing public investment
  mandates do not automatically create citizen dividends.
- EU direct-tax decisions and new own resources normally require unanimity. The
  ordinary EU budget cannot run a deficit. Article 123 TFEU blocks monetary
  financing by the ECB and national central banks.
- Existing EU adjustment and social funds can support retraining and inclusion,
  but do not constitute a permanent EU income entitlement. Member-state tax and
  transfer systems are the faster recurring-income route.
- UK and EU digital competition regimes can require or encourage portability,
  interoperability and data access. Competition reduces lock-in and rents but
  does not automatically distribute income.
- Procurement can impose contract-linked transparency, portability, evaluation,
  skills and transition conditions. EU conditions must generally relate to the
  contract, be verifiable and non-discriminatory. Procurement is not an unlimited
  route to unrelated parent-company equity or domestic preference.
- The EU plans to mobilise €200bn for AI investment, including €20bn for up to
  five gigafactories. Selective national support remains subject to State aid
  rules. The UK is also expanding sovereign compute and AI hardware support.
- Any AI, robot, token or compute levy needs an observable base, import treatment,
  anti-avoidance and interaction with existing taxes. No general UK or EU AI
  levy currently exists.

Official-source URLs and fuller caveats are preserved in
`source/policy_facts.md` for the integrator's later validation.