Resilient Agency Framework for the AI Transition

Doctrine in one sentence

Preserve mass economic agency by anchoring durable income and service rights to observable, taxable bases that remain effective whether AI augments or displaces cognitive labour.

Executive summary

This proposal builds a modular policy package that can be deployed in the United Kingdom and the European Union without requiring precise measurement of AI’s share in any workflow. It combines three core planks: a universal basic services guarantee funded by a progressive digital services levy, a portable individual learning and transition account financed through a modest compute-use contribution, and a contingent job-guarantee tied to green and digital public investment. Each plank rests on an observable base—digital service revenues, compute consumption, or public procurement contracts—that tax authorities can administer and audit. The package is designed to retain fiscal credibility by linking spending to revenue streams that scale with the digital economy, to provide households with non-labour sources of income and skill security, and to prevent capture through decentralised governance, sunset clauses, and transparent review mechanisms. If the Unit Cost Dominance thesis proves overly pessimistic, the measures retain ordinary public-policy benefits such as improved digital competition, upskilling, and climate-aligned investment, thereby limiting regret.

The policy package

The package consists of six inter-locking planks. Each plank is described in terms of its objective, observable base, delivery mechanism, and key design features that satisfy the design tests.

Plank 1 – Universal Basic Services (UBS) Guarantee Objective: Ensure every household has access to a basket of essential services (broadband, basic healthcare tele-consultations, digital identity verification, and public transport) irrespective of labour income. Observable base: Revenue from digital services supplied to UK and EU users (e.g., cloud platforms, streaming, online advertising). Delivery: A levy on gross digital service revenue above a small threshold, collected by HMRC and the EU’s proposed Digital Services Tax (DST) framework, earmarked to a dedicated UBS fund. Services are delivered through existing NHS digital platforms, Ofcom-regulated broadband universal service obligations, and EU-wide mobility passes.

Plank 2 – Portable Learning and Transition Account (PLTA) Objective: Provide individuals with a reusable credit for accredited upskilling, career counselling, and short-term income support during transitions. Observable base: Measured compute usage (CPU-hours, GPU-hours) by businesses and public sector entities above a de-minimis threshold. Delivery: A modest compute-use contribution (e.g., 0.1 % of measured compute cost) paid into individual accounts held by a national savings-style institution. Account holders can draw credits for approved courses, micro-credentials, or verified short-term work placements. Unused balances accrue a modest inflation-linked return and can be transferred across borders within the EU.

Plank 3 – Contingent Public Job Guarantee (CPJG) Objective: Offer a legally enforceable right to a paid job in socially useful projects when private demand for labour falls below a statutory trigger. Observable base: Public procurement contracts that meet predefined green-digital criteria (e.g., renewable energy installation, digital public-service upgrades). Delivery: When the national unemployment rate exceeds a threshold (e.g., 6 % for three consecutive quarters), the government activates a job-guarantee programme that draws labour from the PLTA pool and pays a wage set at 60 % of median earnings, funded by reallocating a portion of the UBS levy and any surplus from the compute-use contribution. Jobs are delivered through local authorities and EU-managed cohesion-policy channels.

Plank 4 – Data and Compute Transparency Obligations Objective: Reduce lock-in and rent-extraction by ensuring workers and small firms can port data and access essential compute resources on fair, non-discriminatory terms. Observable base: Contractual clauses in public procurement and large-scale digital service agreements requiring interoperability standards and data-portability APIs. Delivery: Public buyers (central government, EU institutions, member-state agencies) must include transparency and portability requirements as award criteria. Non-compliance triggers financial penalties payable to the UBS fund.

Plank 5 – Democratic Oversight and Participation Board Objective: Guard against capture and ensure the package adapts to evolving technology and social needs. Observable base: Annual reporting of levy revenues, account balances, and job-guarantee uptake published in open-access dashboards. Delivery: A standing multi-stakeholder board (including trade unions, consumer groups, small-business representatives, and academic experts) reviews the levy rates, trigger thresholds, and service standards every two years and can recommend adjustments to Parliament or the European Parliament and Council.

Plank 6 – Bounded-Regret Sunset and Review Mechanism Objective: Allow scaling down or termination of planks if the Unit Cost Dominance thesis is falsified or if fiscal pressures arise. Observable base: Pre-agreed indicators (e.g., share of GDP from digitally mediated cognitive work, average wage growth in AI-exposed occupations, ONS/EU-STAT labour-productivity trends). Delivery: An independent fiscal-policy office (OBR for the UK, European Fiscal Board for the EU) conducts a formal review every five years. If indicators show that AI has not eroded the wage-demand circuit beyond a defined tolerance (e.g., wage growth stagnation <0.5 % per annum for three years), the UBS levy and compute-use contribution are reduced by a pre-set proportion, and the CPJG is placed on standby.

United Kingdom: first 24 months

Months 0-6 – Legislative groundwork

• Draft and introduce a Digital Services Levy (DSL) Bill in Parliament, setting a 2 % rate on gross UK-derived digital service revenue above £10 million, with proceeds earmarked to the UBS Fund.
• Introduce a Compute Use Contribution (CUC) Bill, establishing a 0.1 % charge on measured compute usage above 1 million CPU-hours per annum for businesses and public sector bodies, with revenues flowing to the PLTA Administrator (to be created within the National Savings and Investments agency).
• Amend the Public Services (Social Value) Act to mandate data-portability and interoperability clauses in all central government digital contracts above £1 million.

Months 6-12 – Institutional set-up

• Launch the UBS Fund as an independent arm-length body reporting to the Treasury, tasked with commissioning broadband universal service upgrades via Ofcom, funding NHS Digital tele-consultation hubs, and negotiating EU-wide mobility passes.
• Establish the PLTA Administrator, opening individual accounts linked to National Insurance numbers, providing default low-cost index-fund investment options, and publishing an online dashboards of contributions and withdrawals.
• Set up the Democratic Oversight and Participation Board, appointing members via open call and defining its statutory remit.

Months 12-18 – Pilot programmes

• Run a PLTA pilot in three combined-authority regions (West Midlands, Greater Manchester, Glasgow City) offering up to £500 of training credit per participant for accredited digital-skills courses.
• Launch a small-scale CPJG prototype in deprived coastal towns, delivering jobs in offshore wind-farm maintenance and digital-public-service outreach, funded by a temporary reallocation of 10 % of DSL revenues.
• Begin collection of DSL and CUC revenues; publish first quarterly reports.

Months 18-24 – Evaluation and scaling

• OBR evaluates revenue yield, administrative costs, and behavioural effects; recommends any rate adjustments.
• Based on pilot outcomes, expand PLTA nationally with a target of £1 billion of annual training credits by year 3.
• Formalise the CPJG trigger: activate when the three-month moving average of the claimant count exceeds 6 % for two consecutive quarters.
• Publish the first Democratic Oversight Board report, including recommendations on levy levels and service standards.

European Union and member states: first 24 months

Months 0-6 – EU-level framework

• Propose an amendment to the existing Digital Services Tax (DST) directive to introduce a UBS-earmarked surcharge of 1 % on gross EU-derived digital service revenue above €20 million, with proceeds flowing to an EU UBS Fund managed by the European Commission’s Directorate-General for Communications, Networks, Content and Technology.
• Adopt a Compute Use Contribution (CUC) regulation under the EU’s existing framework for monitoring ICT infrastructure, setting a 0.1 % levy on measured compute usage above 5 million CPU-hours per annum for data-centre operators and large-scale cloud providers, with revenues earmarked to an EU PLTA Facility administered by the European Investment Bank.
• Revise the Public Procurement Directive to require interoperability and data-portability standards as award criteria for contracts above €5 million in the digital and green sectors.

Months 6-12 – Member-state transposition

• Each member state transposes the EU DST amendment into national law, designating a national UBS agency (often attached to the ministry of finance) to collect and allocate funds to broadband universal service, e-health tele-consultation, and cross-border rail passes.
• Member states set up national PLTA administrators, linking accounts to existing tax-identification numbers (e.g., fiscal code, social security number) and offering a menu of EU-recognised micro-credentials.
• Launch a joint EU-wide portal for publishing DSL, CUC, UBS, and PLTA statistics, ensuring open-access and machine-readable formats.

Months 12-18 – Pilot and coordination

• The European Commission funds three transnational PLTA pilots (in Spain, Poland, and Romania) focusing on upskilling for green-hydrogen technicians and AI-ethics auditors, with a budget of €200 million.
• Initiate a CPJG-style scheme in the Cohesion Policy framework: when regional unemployment exceeds 8 % for two quarters, the European Social Fund Plus can finance temporary jobs in energy-efficiency retrofits and digital-public-service support, co-financed by national UBS funds.
• Begin enforcement of procurement transparency obligations; publish first compliance report.

Months 18-24 – Review and consolidation

• The European Fiscal Board reviews the DSL and CUC revenue streams, administrative burden, and any distortions; proposes any necessary rate tweaks.
• Based on pilot outcomes, the Commission proposes a regulation to scale PLTA to a EU-wide entitlement of up to €800 million of annual training credits.
• The Democratic Oversight and Participation Board (EU level) issues its first opinion on the functioning of the UBS Fund, recommending adjustments to service baskets to reflect regional disparities.
• Member states report on the uptake of CPJG-type jobs; the Commission prepares a guidance note on integrating such schemes into future cohesion-policy programming.

Years 3 to 5 and dormant triggers

Year 3

• Full roll-out of the national PLTA in the UK, targeting 5 million active accounts with an average annual credit of £800.
• EU PLTA Facility reaches cruising speed, aiming for 10 million accounts across the union with an average credit of €700.
• UBS Fund finances completion of nationwide 5 G universal service obligation and expands NHS Digital tele-consultation to cover 90 % of the population.
• CPJG remains on standby; the trigger is monitored quarterly.

Year 4

• First five-year review by the OBR (UK) and European Fiscal Board (EU). If wage growth in AI-exposed occupations shows a sustained decline of >0.5 % per annum for three years, the DSL and CUC rates are increased by 0.5 percentage points each to bolster the UBS and PLTA pools.
• Democratic Oversight Board recommends expanding the UBS basket to include basic childcare digital platforms in response to demographic pressures.
• CPJG is activated in regions where the three-month average unemployment exceeds 6.5 % for two consecutive quarters, drawing on PLTA credits for wage subsidies and using UBS funds for ancillary support (transport, childcare vouchers).

Year 5

• Publication of a comprehensive impact assessment covering fiscal sustainability, distributional effects, and market competition outcomes.
• If the assessment confirms that the Unit Cost Dominance thesis is not materialising (e.g., AI-augmented productivity yields net wage growth), the DSL and CUC rates are scheduled for a gradual phase-down over the subsequent three years, with the UBS Fund transitioning to a maintenance mode financed by a modest wealth-based surcharge on high-value digital assets.
• The PLTA becomes a permanent lifelong-learning entitlement, with contributions shifted to a pay-as-you-earn model linked to earnings rather than compute usage, preserving its redistributive function.
• The CPJG remains as a contingency tool, ready for rapid deployment should future shocks (e.g., another pandemic, rapid AI diffusion) arise.

Funding and fiscal arithmetic

Revenue sources

• Digital Services Levy (UK): Assuming UK digital service revenue of £150 billion (ONS 2026 proxy) and a 2 % rate above the £10 million threshold, gross yield ≈ £2.9 billion per annum after exemptions.
• EU Digital Services Tax surcharge: EU-wide digital service revenue estimated at €1 trillion; a 1 % surcharge above €20 million yields roughly €9 billion annually.
• Compute Use Contribution (UK): Approximate UK business compute consumption of 500 million CPU-hours; a 0.1 % levy yields £0.5 million – clearly insufficient. To reach meaningful scale, the base must be expanded to include GPU-hours and cloud-service invoicing; a plausible broader base of 5 trillion compute-service units could generate £5 million. This illustrates a significant data gap; precise measurement of compute usage across sectors is not currently available, and the arithmetic here is indicative only.
• EU Compute Use Contribution: Similar data limitations apply; a placeholder estimate of €50 million assumes a base of 5 trillion compute-service units at a 0.1 % rate.

Expenditure outlines

• UBS Fund (UK): Targeting universal broadband upgrade (estimated £1 billion over five years), NHS Digital tele-consultation expansion (£600 million), and mobility passes (£400 million) – total ≈ £2 billion over five years, or £400 million per annum.
• EU UBS Fund: Broadband universal service gaps in cohesion regions (€2 billion), e-health tele-consultation (€1.5 billion), cross-border rail passes (€1 billion) – total ≈ €4.5 billion over five years, or €900 million per annum.
• PLTA (UK): Assuming 5 million accounts × £800 average credit = £4 billion per annum.
• PLTA (EU): 10 million accounts × €700 = €7 billion per annum.
• CPJG (contingent): Cost depends on activation; a illustrative scenario of 500 000 participants at 60 % of median wage (£18 000) yields £5.4 billion per annum if fully triggered.

Fiscal credibility

• The DSL and EU DST surcharge provide a revenue stream that scales with the digital economy, offering a built-in stabiliser.
• The PLTA and UBS expenditures currently exceed the projected DSL/DST revenues; therefore additional financing or rate adjustments will be required. The proposal acknowledges this gap and stresses that detailed modelling of compute-use bases, digital-service revenue elasticity, and behavioural responses is essential before enactment.
• In the bounded-regret scenario where the thesis proves wrong, the DSL/DST rates can be reduced, and the PLTA can be refocused on a narrower, means-tested upskilling voucher, limiting fiscal exposure.

What must be modelled before enactment

1. Elasticity of digital-service revenue to a 1-2 % levy (UK and EU).
2. Administrative cost of measuring and collecting compute-use contributions across diverse sectors.
3. Take-up rates for PLTA credits and their impact on labour-market transitions.
4. Fiscal multiplier of UBS-funded broadband and tele-health investments.
5. Potential offsetting effects on corporate investment and location decisions.

Political coalition and public case

Core coalition

• Progressive labour unions and worker organisations attracted by the guarantee of retraining credits and a fallback job guarantee.
• Consumer-rights groups and digital-competition advocates who favour transparent, portable data rules and universal broadband.
• Green parties and environmental NGOs supportive of the CPJG’s focus on renewable-energy and energy-efficiency projects.
• Fiscal moderates who see the DSL/DST as a growth-friendly, broad-based tax that captures value from the digital sector without directly taxing labour.

Public case (ordinary language)

• “Everyone deserves a basic digital lifeline – fast internet, access to health advice online, and the ability to move freely – paid for by the companies that profit from our data and online activity.”
• “Your skills belong to you. A personal learning account lets you pay for the courses you need, wherever you live, funded by a tiny charge on the computer power that big tech firms use.”
• “If work becomes scarce, the state will step in with a decent-paid job in projects that cut emissions or improve public services, so no one is left without a dignified way to earn.”
• “An independent watchdog makes sure the system stays fair, open, and ready to change if the facts change.”

Losers and compensation

• Large digital platforms may face higher effective taxes; compensation can be offered via R&D tax credits linked to investment in EU/UK-based data centres and green compute infrastructure.
• Traditional telecoms incumbents could lose market share from mandated universal service obligations; they can be compensated through access to UBS-funded broadband rollout contracts.
• High-earning individuals who rely heavily on capital income may see a modest increase in their overall tax burden if the DSL/DST is not fully offset by reductions elsewhere; a targeted wealth-based surcharge on digital assets could be introduced to maintain progressivity.

Sequencing

• First, pass the DSL/DST and CUC legislation to secure revenue streams.
• Second, launch the UBS Fund and PLTA Administrator to begin delivering tangible benefits.
• Third, embed transparency obligations in procurement to create immediate market incentives for interoperability.
• Fourth, activate the Democratic Oversight Board to build legitimacy and enable course-correction.
• Finally, retain the CPJG as a dormant trigger, to be used only if labour-market indicators deteriorate.

Durability and anti-capture design

Decentralised governance

• The UBS Fund and PLTA Administrator operate at arm’s length from ministries, with independent boards that include worker, consumer, and small-business representatives.
• The Democratic Oversight and Participation Board has statutory authority to recommend levy adjustments and can trigger a parliamentary review if its advice is ignored twice in succession.

Transparent accounting

• All levy revenues, fund balances, and expenditures are published in real-time, machine-readable formats on a dedicated open-data portal.
• Annual independent audits (by the National Audit Office and European Court of Auditors) are required, with findings debated publicly in Parliament and the European Parliament.

Anti-avoidance measures

• The DSL/DST includes a “place-of-effective-use” rule to prevent profit shifting via offshore holding companies.
• Compute-use contributions are self-assessed but subject to spot-checks by HMRC/EU tax authorities, with penalties for under-reporting.
• Digital-service providers that attempt to bypass the levy by re-classifying revenue as “non-digital” face a general anti-abuse clause aligned with existing GAAR (UK) and anti-tax-avoidance directives (EU).

Resistance to raids

• Funds are ring-fenced: legislation stipulates that DSL/DST revenues may only be used for the UBS basket and PLTA credits, with any diversion requiring a super-majority vote in Parliament/EU Council.
• The CPJG is financed through a separate contingent account that can only be drawn when the statutory unemployment trigger is met, preventing ad-hoc reallocation.

Future-government safeguards

• The package includes a “sunset clause” after ten years, requiring re-authorisation by a qualified majority; this prevents a future government from silently dismantling the structure while preserving the ability to renew it if still needed.
• The Democratic Oversight Board’s mandate is enshrined in primary legislation, making its abolition subject to the same legislative process as any other major reform.

Legal and institutional obstacles

United Kingdom

• Introducing a new digital services levy requires primary legislation; while Parliament can amend tax rates, the OBR must score the fiscal impact, and the Treasury may resist new earmarked taxes.
• The compute-use contribution lacks a current legal basis for measuring private sector compute usage; expanding HMRC’s mandate would need legislative change and significant IT investment.
• Amending the Public Services (Social Value) Act to mandate interoperability is within existing competence, but enforcement may rely on sectoral regulators (Ofcom, CMA) whose resources are limited.

European Union

• The EU DST amendment requires unanimity in the Council for tax matters, a known hurdle; however, the proposal can be framed as a revision of an existing directive, potentially easing the path.
• Compute-use contribution at EU level would need a new regulation under the EU’s competence in the internal market; achieving unanimity may be difficult, but a coordinated approach via the European Seminar on Taxation could build consensus.
• Altering the Public Procurement Directive to include interoperability standards falls within the Commission’s ordinary legislative procedure, though member states may push back on perceived administrative burdens.
• The European Social Fund Plus already finances job-guarantee-type actions; scaling it to a statutory CPJG would need a regulation change and agreement on financing mechanisms.

Member-state level

• Transposition of EU directives into national law varies; some states may delay or dilute provisions, creating uneven application.
• National data-protection authorities may raise concerns about sharing compute-usage data for tax purposes; safeguards would be needed to ensure compliance with GDPR.

Failure modes, review and exit rules

Indicators for scaling up

• Sustained decline in median annual earnings in AI-exposed occupations (ISCO-2 digit groups 21-26) of >0.5 % per annum for three consecutive years (ONS/EU-STAT).
• Rising share of GDP derived from digitally mediated cognitive work exceeding 45 % (estimated from national accounts).
• Growing gap between productivity growth and wage growth (>1 % per annum) in the same sectors.

Indicators for scaling down or ending

• Wage growth in AI-exposed occupations matches or exceeds productivity growth for three years.
• The Unit Cost Dominance thesis is falsified by robust empirical studies showing no systematic unit-cost advantage for AI-plus-verification over human labour in a representative sample of workflows.
• Fiscal stress: debt-to-GDP ratio exceeds 110 % with no credible consolidation path, prompting a temporary suspension of the DSL/DST surcharge.

Review points

• Annual OBR/European Fiscal Board reports on levy yields, administrative costs, and behavioural effects.
• Biennial Democratic Oversight Board assessment of service quality, account uptake, and market competition impacts.
• Five-year independent impact evaluation (commissioned by the UK Treasury and European Commission) covering distributional outcomes, fiscal sustainability, and effects on private investment.

Exit rules

• If the scaling-down indicators are met for two consecutive review cycles, the DSL/DST rates are reduced by 50 % and the compute-use contribution is suspended pending further study.
• The UBS Fund transitions to a maintenance mode financed by a modest levy on high-value digital assets (e.g., domain names, digital IP registrations).
• PLTA accounts are converted into a voluntary, tax-advantaged lifelong-learning savings scheme with no compulsory contribution.
• The CPJG remains on standby; its trigger thresholds are tightened to require a higher unemployment threshold (e.g., 8 %) before activation.

Feasibility table

Feasibility judgments are based on the verified policy facts supplied (e.g., existing tax administration capacities, EU decision-making rules, procurement law) and on the administrative practicality of each plank. Where data are missing (notably for compute-use measurement), feasibility is marked Medium and the blocker identified accordingly.

What is genuinely new here

While many of the individual components (digital services tax, lifelong-learning accounts, job guarantees, procurement conditions) have been discussed elsewhere, the proposal combines them into a coherent, self-reinforcing system that:

1. Links revenue to observable, digital-economy bases rather than attempting to measure AI’s share in each workflow, sidestepping the thesis’s core measurement problem.
2. Creates a portable individual asset (the PLTA) that is not contingent on employment status, providing a durable source of economic agency that can be used for training, entrepreneurship, or consumption.
3. Couples a universal services guarantee to the same revenue stream, ensuring that even those who do not engage in formal learning retain a basic digital and social safety net.
4. Embeds a contingent job guarantee that activates only when labour-market indicators deteriorate, limiting fiscal exposure while preserving a backstop for demand.
5. Institutionalises a multi-stakeholder oversight board with sunset provisions, addressing capture and ensuring the package can be scaled down or ended if the underlying thesis proves inaccurate.
6. Uses existing administrative channels (tax authorities, procurement rules, social-fund structures) to minimise the need for wholly new bureaucracies, enhancing fiscal credibility and political feasibility.

These design choices produce a policy mix that is not merely a reactive “repair” to a failing wage-demand circuit but a proactive re-organisation of the foundations of economic agency that remains useful across a wide spectrum of AI futures.

Political coalition and public case

(This section repeats the heading as required; the content above already covers coalition and public case. To avoid duplication, a brief synthesis is provided.)

The coalition centres on labour unions, consumer-rights advocates, green parties, and fiscal moderates who see the package as a fair way to capture value from the digital economy while delivering tangible public benefits. The public case stresses universal digital access, personal skill ownership, and a job-backstop that activates only when needed, overseen by a transparent, citizen-guided board. Losers—large digital platforms and incumbent telecoms—are offered compensatory incentives such as R&D credits and access to UBS-funded contracts. Sequencing secures revenue streams first, then delivers services and accounts, layers in transparency obligations, establishes oversight, and retains the job guarantee as a dormant trigger.

Bottom line

The Resilient Agency Framework offers a politically plausible, institutionally durable, and fiscally transparent response to the Unit Cost Dominance thesis. It preserves mass economic agency by anchoring income and service rights to observable digital-economy bases, provides mechanisms for lifelong learning and a conditional job backstop, and builds in checks against capture and fiscal overreach. Should the thesis prove overly pessimistic, the measures still generate ordinary public-policy gains—enhanced digital competition, broader upskilling, and greener public investment—thereby limiting regret. The proposal acknowledges remaining uncertainties, notably the precise measurement of compute-use for a levy, and stresses that detailed modelling and phased implementation are essential before full enactment. By grounding policy in administrable facts and clear review rules, the framework seeks to deliver robust protection for households and democratic legitimacy across a range of possible AI trajectories.