Draft plan

Summary

This draft proposes a global programme, phased over about thirty years, to prepare for a world in which machines can do all paid work. Its aim is that every person has a secure income, a home and the essential services they need, keeps a say in how automation is used, and has worthwhile things to do, and that animals, plants and ecosystems recover while this happens.

The plan has seven pillars: economic security, homes and basic services, governance of automation and AI, purpose and participation, planetary stewardship, fairness between countries, and rights and legitimacy. It moves through five phases. The first takes eighteen months and turns this draft into an agreed final plan.

This is a first draft. It lists what the plan could contain and the evidence behind it. It does not yet choose between options, cost them or settle who pays. Those choices are the work of the first phase, set out in the Next steps document.

Why a plan is needed

Automation is advancing, and its end point is uncertain

• The International Monetary Fund estimates that almost 40 per cent of global employment is exposed to AI: about 60 per cent in advanced economies, 40 per cent in emerging market economies and 26 per cent in low-income countries (IMF, 2024).
• The International Labour Organization finds that one in four workers worldwide is in an occupation with some exposure to generative AI, ranging from 11 per cent of employment in low-income countries to 34 per cent in high-income countries (ILO, 2025).
• Employers surveyed by the World Economic Forum expect 170 million new roles to be created and 92 million to be displaced by 2030, and 41 per cent plan to reduce their workforce as AI automates tasks (World Economic Forum, 2025).
• The International AI Safety Report expects general-purpose AI to automate a wide range of cognitive tasks, especially in knowledge work. It finds no effect on overall employment so far, some signs of falling demand for early-career workers in exposed occupations, and disagreement among economists about how large the effects will be (International AI Safety Report, 2026).

The evidence to 2030 describes jobs being disrupted and reshaped. It does not show work disappearing. This plan looks further ahead. It assumes that automation will in time be able to perform substantially all paid tasks, and that the pace cannot be predicted. The plan therefore has to work whether full automation takes twenty years or sixty, and its phases are designed to move with measured progress.

The question is an old one. In 1930 John Maynard Keynes described technological unemployment: unemployment that arises when we find ways to economise on labour faster than we find new uses for it. He thought a fifteen-hour week might put off the problem for a great while, and saw the lasting challenge as how people would use their freedom from pressing economic cares (Keynes, 1930).

The gains may not be shared

Returns from automation go first to the owners of the capital. The IMF expects higher returns to capital to increase wealth inequality, and concludes that the outcome depends on how countries define AI property rights and on their redistributive and fiscal policies. It finds advanced economies more exposed to AI than emerging and low-income economies, and many of the latter less prepared to adopt it (IMF, 2024). Without transfers between countries, the gap in living standards could widen.

The starting point is weak

• Social protection. 52.4 per cent of the world's population is covered by at least one social protection benefit. The other 3.8 billion people have none (ILO, 2024).
• Housing. About 3.4 billion people lack secure, safe and adequate housing, and more than 1 billion live in informal settlements and slums. The global housing deficit grew from 251 million units in 2010 to 288 million in 2023 (UN-Habitat, 2026). Over 300 million people are homeless (UN-Habitat, 2025).
• Nature. Seven of the nine planetary boundaries have been crossed. Six were crossed in the 2023 assessment, and ocean acidification was added in 2025 (Stockholm Resilience Centre, no date).
• Governance of AI. No global framework governs AI. Of the 193 UN Member States, seven are party to all seven of the recent prominent AI governance initiatives and 118 are party to none (UN, 2024a). Risk management for AI remains largely voluntary (International AI Safety Report, 2026).

Vision and principles

The vision is a world in which automated systems do the work that people do not want or need to do, every person's needs are met as of right, people are free to choose how they spend their lives, and the rest of the living world thrives.

Eight principles guide the plan:

• Universal. Entitlements belong to every person as an individual, without conditions.
• No one worse off. Nobody's living standard should fall because of the way the transition is sequenced.
• Shared gains. The returns from automated production are shared widely, within countries and between them.
• People in charge. People decide how automation is used. Automated systems stay safe, accountable and under human control.
• Freedom to choose. People may work, learn, care, create or rest. The plan secures livelihoods. It does not preserve particular jobs or prescribe how to live.
• Within planetary limits. The transition restores nature and stays within planetary boundaries.
• A voice for those who cannot speak. Animals, plants, ecosystems and future generations are represented by named human proxies.
• Evidence first. Pilot, evaluate, then scale, and keep steps reversible where possible.

What the plan would put in place

The measures below are candidates. Each pillar lists what could be put in place and the evidence found so far. Phase 0 will appraise the options and choose.

Economic security

• Universal basic income. A regular, unconditional cash payment to every individual. It starts as a floor and rises as automation takes a larger share of production.
• An automation dividend. A defined share of the returns from automated capital, paid into public wealth funds that pay for the income floor. Alaska offers a precedent: its Permanent Fund was set up in 1976 from mineral revenues and began paying a dividend to residents in 1982 (State of Alaska, no date; Alaska Permanent Fund Corporation, no date). One proposal for firms is a windfall clause: an advance commitment by AI firms to donate a significant share of any extremely large profits from transformative AI (Centre for the Governance of AI, 2020).
• Tax reform. Move the tax base from wages towards automated production, land values and resource rents.
• Global income support. Transfers between countries, so that low-income countries can pay an income floor too.

The evidence so far comes from pilots. A review of 38 basic income experiments in OECD countries found that effects on employment were weak or not statistically significant in most cases, and that most experiments improved wellbeing. None led to a permanent basic income (EPPI Centre, no date). In Kenya, a study covering about 200 villages found that payments did not reduce the total hours people worked, and that a long-term basic income did better than a short-term one on nearly every measure (GiveDirectly, 2023). More than a hundred guaranteed income pilots have been launched in the United States in six years (Jain Family Institute, 2024). Pilots are small and time-limited, so they cannot show what a permanent, universal payment would do to prices, wages or migration. Phase 1 is designed to test this at larger scale.

Homes and universal basic services

• Homes for everyone. An adequate, secure home for every person, with housing treated as a right. Finland has based its homelessness policy on Housing First since 2008, treating permanent housing as a fundamental right and the starting point for recovery. It was for years the only European Union member state where homelessness fell consistently, though the number of homeless people rose by 20 per cent in 2025 (Y-Säätiö, no date). The rise is a reminder that such gains can be reversed.
• More ways to supply homes. Automated, low-carbon construction, retrofit of existing buildings, and rental, cooperative and community-led housing alongside home ownership. UN-Habitat calls for housing to be treated as a human right and for low-carbon, energy-efficient homes (UN-Habitat, 2026).
• Universal basic services. Healthcare, food, water and sanitation, household energy, local transport and digital access, free or nearly free at the point of use. In 2017 the UCL Institute for Global Prosperity proposed universal basic services for the United Kingdom covering shelter, food, transport and information, at an estimated cost of 2.3 per cent of GDP (UCL Institute for Global Prosperity, 2017).

The balance between cash and services is an open question. Cash gives people choice. Services guarantee access where markets fail or prices rise.

Governing automation and AI

• A global automation observatory. Measure which tasks are automated, who is displaced and where the gains go, and publish the results every year.
• Safety and accountability. Standards, certification, liability and redress for automated systems that affect the public.
• A public register. List automated systems that affect people's rights, and assess their impact before they are deployed.
• Human oversight. Keep a person accountable for decisions about rights and entitlements, with a right to an explanation and to appeal.

The programme would build on the recommendations of the UN's High-level Advisory Body on AI: an international scientific panel on AI, a policy dialogue on AI governance, an AI standards exchange, a capacity development network, a global fund for AI, a global AI data framework and an AI office in the UN Secretariat (UN, 2024a).

Purpose, learning and participation

• Lifelong learning. An entitlement to learning at every age, starting with early years education.
• Support for displaced workers. Notice, retraining and income protection while work is still being automated. The IMF regards social protection and retraining for workers exposed to AI as essential in every economy (IMF, 2024).
• Recognised contribution. Recognition and support for care, creative, civic and scientific work, without making any entitlement conditional on it.
• Shorter working time. A shorter working week as a bridge while paid work remains, as Keynes anticipated.
• Participation. Citizens' assemblies on automation, a voice for workers in decisions about their sector, and local institutions for community life.

Planetary stewardship

• Protect and restore. Conserve at least 30 per cent of land, inland water, coastal and marine areas by 2030, as states agreed in the Kunming-Montreal Global Biodiversity Framework (CBD, 2022), and restore degraded ecosystems.
• Automation for nature. Use automated systems for restoration and for monitoring biodiversity.
• A footprint budget. Limit the energy, water, minerals and emissions used by automation infrastructure.
• Animal welfare. Welfare standards for animals in human care, including in automated farming.
• Proxy panels. Standing panels of scientists and knowledge holders who speak for animals, plants, ecosystems and the Earth system in programme decisions.

Fairness between countries

• A just transition finance facility. Mostly grants, so that low- and middle-income countries can join the pilots and build delivery systems.
• Access to technology. Automation technology, skills and infrastructure on fair terms.
• A shared dividend. An international mechanism that passes part of the automation dividend to the lowest-income countries.

Rights, law and legitimacy

• Legal entitlements. Income, housing and essential services become rights that people can enforce.
• An international transition compact. An agreement among adopting states, open to all. The Pact for the Future, adopted at the UN in September 2024 with its Global Digital Compact and Declaration on Future Generations, gives a starting point (UN, 2024b).
• Privacy. Support is never conditional on surveillance, and personal data is kept to a minimum.
• Consent of Indigenous peoples. Free, prior and informed consent for anything that affects their lands, resources or cultures.
• Future generations. An assessment of the effect of major decisions on people not yet born.

How the transition could be phased

The dates are recommendations for planning. They are not forecasts, and later phases would move with the pace of automation measured by the observatory. The Plan lists the steps in each phase.

Who would do what

The programme would be led by a Sponsor and a Programme Board, and run day to day by a Programme Director supported by a PMO Lead, a Design Authority Lead and a Benefits and Assurance Lead. Seven workstream leads cover economic security, basic services, automation and AI governance, purpose, learning and participation, planetary stewardship, engagement and legitimacy, and legal and policy.

Delivery rests with ten types of institution: national governments, regional and local governments, UN bodies, central banks, international financial institutions, industry, civil society and labour organisations, Indigenous peoples' organisations, conservation science bodies and academia and research institutes. The RACI shows who is responsible and accountable for each step.

Main risks

The RAIDD lists 16 risks with their owners and responses. The six rated highest are:

• Political mandate is not won or is lost. Governments decline to adopt the plan, or reverse adoption after a change of government.
• Automation moves at a different pace from the scenarios. Automation arrives much faster, so that people lose income before floors exist, or much slower, so that the cost of floors outruns the dividend.
• Ownership of automated capital concentrates. A few firms or countries own most automated capital, so the dividend available to share is small.
• Inequality between countries widens. Lower-income countries fall further behind because the gains accrue where the capital is owned.
• Automation pushes ecological limits further. The energy, water and mineral demand of automation adds pressure on planetary boundaries that are already crossed.
• Automated systems are unsafe or unaccountable. Automated systems cause harm at scale or operate without meaningful human control.

Open questions

This draft leaves the following questions open for Phase 0:

• How fast will automation spread, and what measurements should trigger each phase?
• What mix of cash and services gives people security and choice?
• How should the returns from automated capital be captured: through tax, public ownership, voluntary commitments by firms, or a combination?
• What level of income is adequate, and who sets it?
• How can low-income countries pay for income floors and services?
• Which institutions should host the programme and the compact?
• What do people want from life when paid work is optional, and what support helps them find it?
• How should the proxies for animals, plants and ecosystems be appointed and held to account?
• How should conflicts between human needs and ecological limits be settled?

What happens next

The Next steps document recommends 13 steps over eighteen months to turn this draft into a final plan. They form Phase 0 of the Plan.

Related documents

The programme documents are listed below. The Control register holds the current version and link for each.

Sources

All sources were read on 10 November 2026.

• Alaska Permanent Fund Corporation (no date). History of the Alaska Permanent Fund. apfc.org
• CBD (2022). Kunming-Montreal Global Biodiversity Framework, Target 3. Convention on Biological Diversity, adopted at the fifteenth meeting of the Conference of the Parties, December 2022. cbd.int
• Centre for the Governance of AI (2020). The Windfall Clause: Distributing the Benefits of AI for the Common Good. 30 January 2020. governance.ai
• EPPI Centre (no date). Basic income experiments in OECD countries: a rapid evidence review. UCL Social Research Institute. Covers experiments completed by November 2021. eppi.ioe.ac.uk
• GiveDirectly (2023). Early findings from the world's largest UBI study. 6 December 2023. givedirectly.org
• ILO (2024). World Social Protection Report 2024-26: Universal social protection for climate action and a just transition. International Labour Organization, 12 September 2024. ilo.org
• ILO (2025). Generative AI and Jobs: A Refined Global Index of Occupational Exposure. International Labour Organization, Working Paper 140. webapps.ilo.org
• IMF (2024). Gen-AI: Artificial Intelligence and the Future of Work. International Monetary Fund, Staff Discussion Note 2024/001, 14 January 2024. imf.org
• International AI Safety Report (2026). International AI Safety Report 2026. Published 3 February 2026. internationalaisafetyreport.org
• Jain Family Institute (2024). Guaranteed Income In The Wild: Summarizing Evidence From Pilot Studies and Implications for Policy. 9 December 2024. jfiresearch.org
• Keynes, J. M. (1930). Economic Possibilities for our Grandchildren. First published in The Nation and Athenaeum, October 1930. economicsnetwork.ac.uk
• State of Alaska (no date). Permanent Fund Dividend Division: historical timeline. pfd.alaska.gov
• Stockholm Resilience Centre (no date). Planetary boundaries. Covers the 2023 assessment (Richardson et al., Earth beyond six of nine planetary boundaries, Science Advances 9, 37) and later updates. stockholmresilience.org
• UCL Institute for Global Prosperity (2017). IGP's Social Prosperity Network publishes the UK's first report on Universal Basic Services. 11 October 2017. ucl.ac.uk
• UN (2024a). Governing AI for Humanity: final report of the Secretary-General's High-level Advisory Body on Artificial Intelligence. United Nations press release, 19 September 2024. un.org
• UN (2024b). Pact for the Future, with its annexes the Global Digital Compact and the Declaration on Future Generations. United Nations, adopted 22 September 2024 (A/RES/79/1). un.org
• UN-Habitat (2025). Strategic Plan 2026-2029. unhabitat.org
• UN-Habitat (2026). World Cities Report 2026: The Global Housing Crisis: Pathways to Action. Launched 19 May 2026. Read as reported by the IISD SDG Knowledge Hub. sdg.iisd.org
• World Economic Forum (2025). Future of Jobs Report 2025. Press release, 8 January 2025. weforum.org
• Y-Säätiö (no date). Homelessness in Finland. Y-Foundation. ysaatio.fi