Epistemica

The anticipated result of the proliferation of ai—that is, the imagined world that is expected to result from ai, from its automations and worker replacement— is strangely unimaginative. We keep roughly the same schools, hospitals, neighborhoods and institutions, but companies need fewer people to operate them. Positions for workers disappear as everything else stays recognizably the same so the argument tends to reduce to how many jobs survive and how many people will be un- or under-employed. The main problem with these common imaginings is that they tend to overlook how profoundly unfinished our society remains, and how much more we could accomplish during and/or after the proliferation of ai.

Compared to what a decent society could look like, the US remains enormously underdeveloped. We have the clear capacities to pursue far better education, healthier communities, richer cultural lives, and more reliable public systems, which are just some among dozens and likely hundreds more goals. An economy that cannot connect people willing to contribute with those unmet needs has glaring issues.

It should be obvious now that prior technological advancement or revolutions that threatened to upend the role of humans in work resulting in massive and often expedient adjustment. Krugman’s “Technology and Jobs” (2025) and Autor’s “AI Could Actually Help Rebuild the Middle Class” (2024) examine earlier technological upheavals and the reorganization of work. Economies have repeatedly adjusted, even if painfully and unevenly. But I want to focus instead on the untapped potential in the US (and elsewhere of course) to improve society directly by human action.

Even if some defense of ai is assumed here, the intention is not a broad defense of the use of ai. Its usefulness must be demonstrated in particular settings, and its threats of course deserve serious attention. But the changes to the economy and the ways in which people work are already happening, and it is worth considering what humans can and should focus on during this time.

The International Labour Organization’s 2025 assessment held that job transformation the likelier overall effect of generative ai, because occupations combine activities that still require people. Even so, reduced labor requirements could expand our capacity to address neglected needs.

Consider first the physical makeup of the country. The American Society of Civil Engineers awarded American infrastructure a C in 2025, an improvement from prior years, but nine categories remained in the D range. Even repairing existing systems would demand sustained effort. We could also build communities where reliable transit, clean water, accessible buildings and affordable broadband are ordinary expectations. Better electricity networks and protection against floods and extreme heat would enlarge what people can safely do.

Healthcare deficiencies are even more obvious in the US. Federal workforce projections anticipate shortages across many health professions by 2038, with particularly severe gaps outside metropolitan areas.  We should be developing accessible primary care, mental healthcare, rehabilitation and dependable support for aging and disabled people. A system that saves clinicians paperwork could give them more time with patients.

Development also involves setting higher goals for technology itself. We could design equipment that is easier to repair, homes that need less energy and public services that people with disabilities can use independently. The ambition should be to expand human capability while also reducing needless waste. There is no reason to assume that today’s arrangements represent the best we can achieve simply because an algorithm can operate parts of them more cheaply.

Scientific ambition should be focused as well. We could support more basic research, investigate neglected diseases, improve public laboratories and make established knowledge easier to use. Discovery also requires education, equipment, careful experiments and institutions willing to support uncertain results.

Of all the deficiencies in society today, education is among the most important and may actually be the prerequisite for many others. Yet we make teaching a costly and daunting choice for people we need. The Economic Policy Institute estimates that public school teachers earned 25.2% less in weekly wages than comparable college graduates in 2025. Better benefits narrowed, but did not erase, that disadvantage.  RAND’s 2026 survey also found teachers more likely than similar working adults to report poor well-being on every measured indicator.  A person can want deeply to become an effective teacher and reasonably conclude that the career will not support a decent life. Competitive pay, manageable workloads, classroom support and professional autonomy would make the aspiration more practicable. These are all goals that we can work towards. These is plenty of blame to go around in politics for the poor education system in the US. Governor Abbott loves to cut public school funding as he allows private schools, which are now partially funded by taxpayers, to dump poor-performing students into public schools just before testing. But even beyond politics, there is so much more to be done to improve education in the US.

Our communications systems also have room for enormous improvement. We need dependable local reporting, public information people can understand, accessible services and places to deliberate without every interaction becoming an opportunity for advertising or surveillance. Editors, librarians, translators and educators help people interpret information and judge competing claims.

The arts are owed development for their contributions to a better society. A richer society would give more people time to learn an instrument, write seriously, perform, make films or encounter work beyond whatever commands immediate commercial attention. Libraries, theaters and affordable studios could widen participation. Artistic development includes the growth of the person making something, and the relationships that form around it.

Beyond employment, there is the ordinary experience of being alive. We could make neighborhoods less isolating, public spaces more welcoming and daily routines less exhausting. People could spend more time with their children, care for relatives without financial collapse and participate in local decisions. These are substantial improvements in human welfare, even when they produce nothing that can be sold, which itself reflects a cultural deficiency that may only be addressable via improved education.

Internationally, the unfinished work is greater still, and the severing of critical USAID programs has added damage to existing needs. A September 2026 inspector-general advisory confirmed that nearly all non-statutory USAID positions had been eliminated and all its overseas missions closed, while many functions moved to the State Department. Lost institutional knowledge and working relationships take sustained human effort to rebuild.

. In a study published in July, 166 partners implementing the U.S. HIV program PEPFAR reported 1,714 service sites closing after funding was terminated or payments delayed. Congress approved substantial U.S. HIV funding in February 2026.  Rebuilding dependable services nevertheless requires sustained support for local professionals and community organizations. Countries facing disease, hunger and displacement need stronger capacity and reliable partners.

All of this collectively exposes the weakness in the claim that society might soon have little use for people. There is ample useful activity available and if ai frees up more time to focus on the clear societal deficits, we should actually focus on them (and if not, we should anyway).

Step one might be to top electing billionaires. We should elect people committed to building public capacity and giving workers a meaningful say, which of course requires changing the rules through which wealth buys political influence.

Democratic purpose should shape investment decisions instead of the wealthy writing the laws and reinforcing their own influence. Communities should help identify the services they lack and evaluate whether new spending improves daily life. Workers should participate before employers impose systems that change their duties. Public institutions need the expertise to scrutinize vendors and the independence to reject bad proposals. Contracts and spending decisions should be publicly accessible. We should judge development by the capabilities people gain and the harms they escape.

Economic restructuring should make socially valuable work financially viable. Governments should fund durable careers in education, care, infrastructure, research and the arts, with clear public goals and competent oversight. Training should include paid apprenticeships and support for living expenses. People cannot move into needed professions through encouragement alone. Projects also require materials, energy and skilled supervision, and as such investment must expand real capacity alongside funding.

The gains from higher productivity should help pay for that development. Taxing concentrated wealth and capital income more effectively, enforcing competition rules and strengthening collective bargaining would help distribute resources and decision-making power. Public investment should secure public benefits. Worker ownership and representation can give employees greater influence over whether technology improves their jobs, intensifies workloads or simply eliminates them.

There is also a case to be made for share gains through shorter working time where productivity permits, with pay protected. (I can hear the shouts of “socialism!” now.) Healthcare and basic security should remain even during a change of employer. Greater freedom could support study, family life, artistic practice and civic participation. A decent society should allow people to flourish without requiring every hour released by a machine to become another hour sold. Such freedom should be available across income levels and occupational backgrounds.

Nor should we wait for automation to provide the opportunities commitments. The need for good schools and accessible care exists now. If ai delivers less than promised, those investments are still easily worth it. If it delivers substantial savings, the case for making them becomes stronger. Our obligations to one another do not depend on a technology company’s forecast, but with the trajectory of ai right now, it is definitely worth realizing that humans shouldn’t just click buttons on screens all day.

We have much more to do. Our politics and economy should finally be organized to let us do it.