In brief

What this examines

If AI is going to change everything, the essentials of a decent human life deserve a central place in the plan. Food, clothing, shelter and education should be reliably available to everyone. That is a minimum obligation; it does not make unjustified risks acceptable.

Why it matters

The debate over superintelligence needs to address both safety and purpose. Technical progress alone cannot ensure that its benefits reach people whose basic needs remain insecure.

Key ideas

  • Direct AI toward secure access to food, clothing, shelter and education, alongside climate action and human freedom.
  • Assess public benefit through lasting improvements in people’s lives, with affected communities involved in defining success.
  • Preserve enterprise and productive risk-taking while questioning whether deprivation must remain a condition of economic participation.
  • Make commitments global, proportionate and accountable; a worthy destination cannot justify an unsafe development path.

Take the promise seriously

My challenge is to take such ambitions seriously enough to ask how they become commitments. Who receives the benefits? What resources are attached? How would we know whether an initiative was improving lives, and what would change if it failed?

The strongest invitation to the industry is to bring its ingenuity to that task. Treat the difficulty of delivering abundance as a problem worthy of exceptional people. Give the relationship between technical capability and public benefit the attention currently given to making the technology more capable.

Start with the life we want

The future I keep returning to resembles the social aspiration of Star Trek: material security, room for curiosity, and a human identity larger than the country of one’s birth. Its appeal is the freedom to learn, explore and contribute when survival no longer depends on a paycheque. The franchise’s own discussion of its economic vision makes that connection between abundance and individual freedom explicit.

For me, the starting point is straightforward. Access to a decent life should not depend on whether the labour market currently values what a person can sell. People should be able to care for relatives, study, create, conduct research or build something new without their basic needs becoming collateral for the decision.

Make abundance tangible

Food is a useful place to begin. An AI initiative could investigate better crop planning, reduced spoilage or more reliable distribution. Those are possible avenues, not a claim that one model can end hunger. A meaningful project would need farmers, logistics operators and local communities to determine where assistance is useful. Its success would depend on whether people actually became more food-secure.

Housing invites the same discipline. We could direct technical effort toward construction methods, materials, energy efficiency and planning. But a cheaper design does not necessarily become an affordable home. Land, financing, infrastructure and the way savings are passed on all matter. The goal should reach as far as the household receiving the keys.

In education, imagine assistance shaped around a learner’s language, circumstances and goals, with teachers involved in its design and evaluation. Ask whether understanding improves, whether errors are caught, and whether the service remains accessible to families with little money. Access to a chatbot is a beginning, not an education in itself.

Freedom is an economic ambition

There is an argument here that takes individual freedom seriously. A person with secure housing and enough to eat may have more room to start a business, retrain, leave an exploitative situation or attempt work that does not pay immediately. Material security could widen the pool of people able to take a productive risk. It need not imply that ambition disappears.

I am not proposing that every difference in reward be abolished. I am questioning the necessity of deprivation as the price of an economy that rewards achievement. We can value invention while asking whether the gains from extraordinary productivity should establish a more generous minimum for everyone.

Humanity is larger than the home market

Governments have responsibilities to their citizens, and security concerns cannot simply be wished away. But a claim to benefit humanity must extend beyond the countries hosting the leading laboratories. A child’s entitlement to food or education should not turn on whether their country is an important customer, investor or strategic ally.

Global benefit also means giving affected people a say. A community should be able to identify the problem, challenge the proposed intervention and assess the result. Being handed a tool designed elsewhere is different from having meaningful influence over its purpose. Participation has to reach further than a launch photograph or a consultation after the major decisions are made.

An obligation worth discussing

If we are considering legal obligations for frontier AI companies, helping advance these global initiatives deserves a place in the discussion. I would support a proportionate obligation even where fulfilling it reduced potential profits. A commitment that applies only when it costs nothing is a fragile basis for a universal promise.

One possible starting point would be funded plans addressing basic needs, developed with affected countries and communities. They could identify resources, partners, milestones and a means of independently assessing progress. The details should be contestable. We would need to distinguish effort from results and honest failure from an initiative designed chiefly to satisfy a reporting requirement.

We can ask for more

There is a temptation to treat any positive vision as a justification for accepting the danger required to reach it. I reject that bargain. An attractive account of universal abundance does not establish that a particular development path is responsible. Safety needs its own evidence and justification, regardless of how inspiring the destination sounds.

I do not have the blueprint for a world of abundance and freedom. I am unwilling to accept that we need one before we can ask more of this race. If AI can change everything, helping humanity secure the essentials of a decent life is a reasonable place to begin.

Selected primary sources

Open the primary-source pages used to verify the claims summarized here.

Read on Substack Substack edition