Deutsche Fassung: Der goldene Käfig der Post-Arbeit

AI 2040: Plan A imagines a future that feels almost suspiciously reasonable within the current AI debate. The international race toward superintelligence is slowed down and made verifiable. AI research becomes more publicly organized. Several providers remain capable of acting, rather than one company setting both the technological and political direction. A substantial share of the new wealth is returned to the population through a citizens‘ dividend.

The authors do not present this as a confident forecast. They describe it as a recommendation and a scenario stress test. That is what makes it useful. The text asks not only what may become technically possible, but which institutions would be needed to make a positive development durable.

This is already more serious than the familiar future in which a handful of companies win the race and everyone else eventually receives the updated terms of service. Safety, ownership, international oversight and distribution are treated as one political problem. That is progress.

From a New Work perspective, however, the decisive question begins precisely where the scenario appears to have reached its goal.

A Dividend Protects People from Need, Not from Meaninglessness

A citizens‘ dividend can create material security. It can reduce the pressure to accept any available job. It can shift bargaining power and release time. In a society where machines perform a large share of cognitive and physical labor, none of this would be minor.

But income alone does not tell people what they are able to do with their freedom.

Having money does not give someone a workshop. Having time does not create a place to learn. No longer having to work does not automatically provide a task, a community or a voice in the systems that structure everyday life. A million dollars can release a person from many dependencies. It does not teach them how to repair a solar installation, accompany a child, negotiate a conflict, organize a town or challenge a machine-made decision.

A very well-funded waiting room is still a waiting room. The refreshments are better. The departure board is still missing.

This is not an argument against material security. On the contrary, without security, freedom often remains a privilege. But security is a condition of freedom, not its complete form.

Post-Work Can Still Mean Heteronomy

The story of liberation through automation has a blind spot. It quietly assumes that when necessary labor disappears, dependence disappears with it.

That does not have to follow.

A society may require very little human wage labor and still be highly centralized. A small number of infrastructures could decide how energy, compute, information, mobility, health and access are distributed. People might receive a generous income while remaining users of systems whose rules they neither understand nor can change.

The machine performs the work. The institution defines the frame. The person may choose, as long as the choice has already been provided.

This would no longer be a traditional work society. It would not yet be a free society either. It would be a golden cage of post-work: comfortable, efficient and remarkably difficult to rebuild.

New Work therefore asks more than how people make a living. It asks what allows them to grow, what they can really, really want, and which means they possess to turn that desire into reality. Frithjof Bergmann’s idea of freedom was never merely leisure. It concerned the ability to begin something of one’s own and help shape the conditions of one’s life.

Freedom Needs Substrates

Freedom remains abstract when it has no places, tools or access. A credible post-work society therefore needs freedom substrates.

These include open places of learning where people do not merely operate systems but understand relationships and consequences. Workshops and local production facilities where ideas can become material. Access to energy, data and compute without every action depending on one provider. Spaces for care, craft, culture and community production. Institutions that do not treat dissent as system noise, but as necessary feedback.

Even a highly advanced AI remains a tool within a political architecture. If it communicates knowledge, it must remain possible to trace which sources, objectives and omissions shaped its answers. If it prepares decisions, people and procedures must be able to object, correct and assign responsibility. If it creates prosperity, society must distribute more than money. The capacity to understand and shape the productive systems must not disappear.

A dividend distributes returns. Freedom substrates distribute the capacity to act.

That is the difference between provision and enablement.

A New Work Test for a World Without Compulsory Labor

A positive post-work scenario should face at least five questions.

First, are people materially secure without being placed inside new digital dependencies?

Second, do knowledge, productive tools and infrastructure remain accessible, or do they become concentrated in a few institutions despite the broad dividend?

Third, are there real spaces of learning and experience where judgment, capability and responsibility can grow?

Fourth, can people effectively contest technological and institutional decisions?

Fifth, do new forms of community production and care emerge, or does the population primarily become a comfortably supplied audience?

These questions are not arguments against automation. Many pointless, dangerous and exhausting tasks should disappear. Nobody needs to defend the manual reconciliation of three contradictory spreadsheets as the final stronghold of human dignity. Some office processes were historical misunderstandings with calendar access long before AI arrived.

The task is not to preserve bad work. The task is to use the released time and productivity to make better forms of human activity possible.

From Distribution to a Democratic Building Project

The strength of AI 2040 is that it does not treat power and distribution as a social appendix to be added later. International verification, public research and several capable actors form part of the scenario’s core. It makes one point clear: a positive AI future will not emerge automatically from more capable models. It has to be built institutionally.

The New Work addition is that this building project continues even after wealth has been distributed well.

Public AI research requires public accountability. A citizens‘ dividend requires access to education, energy and production. A world with less wage labor needs places where people do not merely occupy themselves, but learn, contribute and assume responsibility together. Technical safety requires a society capable of dissent.

This sounds less spectacular than superintelligence. It consists of workshops, schools, libraries, municipalities, open technical infrastructure and well-designed procedures. Future studies occasionally struggle with things that require a caretaker’s key. Society generally does not.

The Open Question Behind Plan A

AI 2040 offers a valuable alternative to an uncontrolled race. It shows that slowing down, verification, public research and broad distribution can be considered together. Its scenario deserves a generous and serious discussion.

That is precisely why the remaining question should not disappear:

What does a society do with the freedom it has gained?

If people are merely released from work without receiving the means to shape their world, freedom remains passive. If they gain access to knowledge, tools, communities and contestable institutions, post-work can become New Work: activity no longer determined solely by systemic compulsion, but by human curiosity, responsibility and the desire to bring something meaningful into the world.

The future will therefore not be decided only by how much machines produce or how fairly the returns are distributed. It will be decided by whether people can still begin, change and take responsibility for something together.

Liberation from work may be a beginning. Developing the capacity for freedom is the real task.

Sources and Further Reading

• AI Futures Project: AI 2040: Plan A, recommendation and scenario stress test: https://ai-2040.com/
• Flowbook AI: trust architecture, freedom substrates and New Work in the age of AI: https://flowbook.newwork-newculture.dev/en/

Photo: Thomas Schneider / code intuitif