Economy & democracy
Jobs and education
Start with what automation can do. Then ask whether a person makes the experience better for the people using it.
A job title does not establish a role for human work. Assess the specific function, its automated alternative, and the extra value a person brings to its users. Event scheduling and bookings belong in the automation baseline; a human host is a different contribution to assess.
For paid work in the POE, the test is whether human involvement improves a valued outcome or experience for its recipients. How much it improves, for whom, and at what cost determines the case for choosing it. Public priorities and funding remain democratic choices. People’s democratic right to participate is a separate principle; it does not depend on this economic test.
Compare a specific function
Compare the function
Where does a person add value?
Each example names the customer benefit that justifies a person’s role in the company. The assessment checks whether it is delivered, for whom, and by how much. Customers here include people using publicly funded services.
What machines can cover
AI can plan schedules, coordinate bookings, send invitations, and optimise resources. The scenario assumes these functions can be automated; real deployments still need their capabilities and reliability checked.
What customers gain from a person
No added benefit from scheduling alone. Customers need accurate bookings, suitable schedules, and quick changes. When automation delivers these equally well at equal or lower cost, employing a person to repeat the task adds no customer value. A personal welcome and a shared occasion belong to the separate hosting function.
How to measure the customer benefitCompare errors, successful bookings, response time, cost, and organisers’ satisfaction for the same event requirements.
What education should developRelevant training must improve the assessed outcome. A qualification alone does not establish an advantage over automation.
Your assumptions for this example
The starting numbers are invented to demonstrate the comparison. Replace them with evidence from the service and people being considered.
This is technical capability. People can still choose human involvement in work that can be automated.
No added consumer quality in this example.
The difference is in rating points, not a measured percentage improvement. A preference or quality gain does not by itself establish how many people will buy the service.
Two reasons a person may do the work
A technical gap: some part of a defined service cannot yet be automated to the required standard. This may create human work today, but the gap can change as technology improves.
A chosen human contribution: automation can deliver the baseline service, but its recipients get a better outcome or experience when a person is involved. This is the defining case for personal opportunity in the proposed framework.
Human relationships, participation, authorship, and watching people perform can be part of the quality people value. The benefit must be assessed for the relevant users; it should not be assumed from a profession’s name or training requirements.
How much better and for whom
Define the users and the function first. Compare the best suitable automated service with a version containing a precisely described human contribution. Both versions can use technology.
Measure the outcomes users care about: for example learning progress, reliability, enjoyment, connection, accessibility, or the value of human authorship. Assess cost and the time required alongside quality. Where possible, compare functional quality without provider labels, then test preferences with human or automated provenance disclosed.
Report differences between people as well as the average. A human contribution can improve some users’ experience, add little for others, or make it worse. Customer preference, ability to pay, and democratically funded public benefit are separate considerations.
A paid role needs valued human contribution, enough demand or an approved public purpose, sustainable payment, and people who want and are prepared to do it. The workforce explorer lets you vary those assumptions.
Science belongs in both economies
Automation advances science. People can help each other understand it. In the highly automated economy considered here, most scientific production is assumed to be automated. A personal contribution to investigate is helping people understand findings, discuss evidence, ask questions, and connect science to their own lives.
Examples include a knowledgeable person leading a science workshop, discussing a museum demonstration, or helping a community understand research relevant to a public decision. AI can also explain science and answer questions. Compare whether involving a person improves understanding or the experience for the particular audience, with AI available in both versions.
Measure understanding, not just confidence. Can people explain the finding accurately, apply it, recognise its uncertainty, and remember it later? Human interaction may help some audiences more than others. Scientific research itself can also involve people where they demonstrably improve its outcome; neither a job title nor a doctorate establishes that advantage.
Research does not need an immediate paying customer to have value. Democracy can fund basic research for public and future benefit, including highly automated research. Public funding alone does not make it POE work. Under this proposal, funding for personal scientific work goes to the organisations delivering it, which pay the people doing the work.
A research company can combine automated experiments with people who help others understand the results. Universities, public institutes, and nonprofit organisations can also be providers. Funding needs to distinguish scientific output from public understanding and participation, and allow appropriate time for uncertain research.
Democracy sets budgets, broad priorities, and accountable funding rules. Scientific conclusions depend on evidence, reproducibility, and open criticism. Research programmes should protect freedom of inquiry, including findings that challenge current political priorities.
Explaining science well can require years of scientific education. A person needs to understand the methods, evidence, and limits, alongside learning how to listen and communicate. POE can support that demanding path when people choose it and their contribution is valued. Training readiness in the explorer represents that preparation; moving the slider does not make it happen instantly.
The explorer includes scientific research and discovery and understanding and explaining science. The first starts with no assumed human quality advantage; the second illustrates a possible benefit from personal discussion and explanation. Both are editable scenarios, not findings about employment or capability.
Politics is people shaping society
AI can support political work. Democratic authority comes from people. In this framework, people discuss what matters, represent one another, negotiate disagreements, choose priorities, and hold power accountable. The right to have a say remains with people even when AI can produce a better analysis.
Political participation can also create valued personal work. Examples include facilitating public discussions, community organising, helping people express their concerns, and supporting dialogue between residents and representatives. AI can help with preparation and analysis; assess whether personal involvement improves participation, understanding, and the experience of being heard.
The politics example in the workforce explorer counts paid participation support. It does not count every voter as a worker, allocate elected offices, or make political rights depend on employment. Its automation and quality sliders apply to the support service, not to who has the right to decide.
Under approved POE programmes, companies or civic organisations can receive funding for this work and pay the people delivering it. Proposed funding rules should be transparent and open to different political views, with people free to criticise the funder. Success means meaningful participation and accountability, rather than agreement with those in power.
Education can improve political work; a political voice does not require a qualification. Listening, facilitation, negotiation, understanding institutions, and specialist policy knowledge can take sustained learning. Everyone’s freedom to discuss, remember, and think belongs at the foundation. Read the democracy framework.
Examples to investigate
Each example states the benefit a human role needs to deliver to customers. Its size and value depend on the people using the service. The measurement criteria below each example establish whether the company delivers that benefit.
Event scheduling and logistics
Automation baseline. AI can plan schedules, coordinate bookings, send invitations, and optimise resources. The scenario assumes these functions can be automated; real deployments still need their capabilities and reliability checked.
Value to customers: No added benefit from scheduling alone. Customers need accurate bookings, suitable schedules, and quick changes. When automation delivers these equally well at equal or lower cost, employing a person to repeat the task adds no customer value. A personal welcome and a shared occasion belong to the separate hosting function.
How to measure the benefit. Compare errors, successful bookings, response time, cost, and organisers’ satisfaction for the same event requirements.
Hosting a shared experience
Automation baseline. AI can organise an event, suggest activities, and support delivery. Those planning tasks do not require a person simply because the event involves people.
Value to customers: A personal welcome and a sense of belonging. Participants share an occasion with a host who welcomes them, introduces people, and brings them into the activity. The customer benefit is feeling included and connected to the people sharing the experience.
How to measure the benefit. Ask participants to compare inclusion, enjoyment, connection, and willingness to choose the human-hosted version at its actual price.
Coaching and learning practice
Automation baseline. AI can explain subjects, generate exercises, adapt practice, and provide feedback. Expertise or teaching credentials alone do not make these tasks immune to automation.
Value to customers: Motivation and progress through a working relationship. Learners practise with a coach they choose, work towards shared goals, and receive encouragement from someone who knows their progress. The customer benefit is sustained engagement that helps them develop and retain skills.
How to measure the benefit. Compare skill retention, progress on agreed goals, attendance, motivation, accessibility, and the learner’s experience.
Conversation and personal support
Automation baseline. AI can converse, provide information, organise support, and assist with routine tasks. The technical service should be assessed separately from the relationship.
Value to customers: A reciprocal relationship with another person. Recipients spend time with someone who knows them, listens, follows up, and shares the experience. The service provides human companionship and continuity alongside practical support, for people who value that relationship.
How to measure the benefit. Let recipients define valued outcomes; compare continuity, reported connection, practical benefit, accessibility, and their choice of provider.
Craft making and shared practice
Automation baseline. AI and robotics can design and manufacture objects. Quality, complexity, or many years of human study do not by themselves establish a technical limit to automation.
Value to customers: A connection to the maker and the making. Customers buy the work of a particular person, discuss its creation, or take part in making it. Human authorship, the relationship with the maker, and the experience of making something together are part of what they choose, alongside the object's practical quality.
How to measure the benefit. Compare usability and durability, then separately test informed preferences for provenance, collaboration, and price.
Human sport and live performance
Automation baseline. AI can generate entertainment, analyse performance, run administration, and create automated alternatives. Many supporting tasks can be automated.
Value to customers: The experience of human effort and achievement. Audiences watch people develop and demonstrate skill, compete, perform, and respond to one another. Following those people and sharing the occasion are part of the experience customers choose.
How to measure the benefit. Compare audience choice, participation, enjoyment, repeat attendance, and the amount people or public programmes are willing and able to pay.
Scientific research and discovery
Automation baseline. AI and robotics can automate parts of scientific work, including experimental planning, execution, and analysis in defined settings. This scenario starts with a fully automated research function as a comparison, not a claim that all science is already automated.
Value to customers: No added research benefit is assumed in this example. Users of research need reliable, useful knowledge. A human role in the research company needs to improve those findings, their relevance, or the cost of producing them. The starting scenario assumes automation delivers the research equally well; the separate science-explanation function describes value for people learning about the results.
How to measure the benefit. Assess validity, reproducibility, relevance, uncertainty, cost, and longer-term knowledge gained. Set appropriate review periods for uncertain research. Assess participation and learning separately from scientific output.
Understanding and explaining science
Automation baseline. AI can explain findings, answer questions, create demonstrations, and guide learning. In this scenario, most scientific production is automated. Assess automated explanation too before choosing a human role.
Value to customers: Understanding through personal discussion. Participants explore findings with a knowledgeable person, ask questions, discuss the evidence, and connect it to their lives. The customer benefit is understanding they can use and explain, together with the experience of learning through a discussion with another person.
How to measure the benefit. Compare what people can accurately explain afterwards, how they apply the knowledge, whether they recognise uncertainty, and what they remember later. Also assess participation and their experience. Greater confidence or trust alone does not establish better understanding.
Political discussion and community organising
Automation baseline. This scenario treats preparation and participation support as automatable: gathering information, comparing policy options, preparing meeting materials, and organising discussions. Accuracy, bias, and accessibility still need assessment. The comparison concerns support services; it does not replace citizens or give AI political authority.
Value to customers: Being heard and shaping shared decisions. Participants have a forum to express concerns, hear different views, and work together on decisions that affect them. A paid organiser or facilitator provides personal support for that participation and helps concerns reach representatives accurately. The benefit is a more effective voice; the right to participate remains independent of this paid service.
How to measure the benefit. Assess whose voices are heard, understanding of alternatives, accurate representation of concerns, accessibility, freedom to disagree, and whether people can challenge decisions. Agreement with the organiser or funder is not the measure of success. These scores assess the support service, not the worth of people's political views or their right to participate.
Education and the time to develop expertise
The POE must make room for years of education, apprenticeship, and practice. Reskilling means giving people a real path into new work as the economy changes. That path can include a full professional education and sustained practical training. People should also have opportunities to deepen an existing skill throughout their working lives.
Long study and practice can build rare expertise. The economic value lies in the skill, judgement, quality, and trust developed through that effort. Where these are valued and scarce, people can earn more. The POE should give people reasons to pursue mastery and clear opportunities to progress from learner to experienced practitioner or specialist.
People should be able to undertake demanding education without needing wealthy parents. Democratically chosen support can cover education costs, living costs during study, childcare, and accessibility needs. Paid apprenticeships, supervised placements, and flexible study can make these paths usable at different stages of life. Publicly supported programmes should have clear standards, suitable support, and meaningful assessment of progress.
Education and paid opportunities need to be planned together. The education paths and forms of support described here are proposals that people can accept, change, or reject through democratic decisions. Providers should explain the work a course prepares people for and who is likely to pay for it: customers, employers, or publicly funded services and projects. As tools and needs change, people should be able to update their learning and build on the expertise they have already earned.
Different incomes and rewards
A good life for everyone, with room to earn more. People in the POE earn different incomes. Effort, learned skills, natural talent, responsibility, and enterprise can bring greater rewards and more choices. Some people are fortunate to have exceptional abilities; everyone should have the chance to discover and develop what they can do. A good life should be within reach for people with ordinary abilities too.
Abundant automated production can coexist with scarce experiences. The Eiffel Tower cannot accommodate everyone whenever they want to visit; particular places, live events, and people's time have limited capacity. Higher earnings can give people more choices about the goods and experiences they can afford.
Additional earnings should improve a person's circumstances. Taxes and the withdrawal of support should be designed so that taking on paid work or earning more remains worthwhile.
Evidence and the limits of this model
The ILO’s 2025 occupational exposure study builds its analysis from tasks, expert assessment, and model predictions. That supports looking inside occupations. It does not establish the quality gains, robotic capabilities, or workforce numbers in these examples.
For a concrete example of research automation, the A-Lab study in Nature describes a system that plans, performs, and interprets materials-synthesis experiments using machine learning and robotics. It is evidence about a defined experimental setting, not all scientific work. UNESCO’s Open Science Recommendation supports wider participation in science. Neither source establishes that human involvement always improves outcomes or supports the numerical assumptions in this explorer.
The percentages and quality scores in the interactive tools are illustrative assumptions. They are not measured automation rates, current employment statistics, or predictions. Relevant service trials, consumer research, skills data, and costed budgets are needed before making real programme decisions.