AI, Money, and the End of Productivity as a Life Goal

I asked my AI assistant a question that had been bothering me: if machines eventually become better at making money than we are, what happens to a life organised around making money?

I started with software. If AI keeps lowering the cost of building it, perhaps the same thing happens to other forms of expensive expertise: analysis, management, capital allocation. Eventually, some of the decisions that justify enormous salaries and extraordinary concentrations of power.

Would our obsession with money, time, and productivity start collapsing under the weight of the tools we’d created to maximise them?

The answer I’ve arrived at is more uncomfortable than a prediction of abundance. AI can make certain kinds of work cheaper while leaving the people who own the systems more powerful. Money can remain necessary even as earning it becomes a less convincing explanation for how we spend our lives.

That’s the tension I want to stay with.

Will AI make money less important?

AI doesn’t automatically make money obsolete. It can change which activities earn money, who captures the returns, and how much human labour an outcome requires. Ownership, scarcity, and the rules governing access still matter.

Imagine an AI that consistently allocates capital better than a human executive. That’s a hypothetical capability, rather than something I’m claiming has been established across the messy conditions of running a company.

Even in that scenario, the owner can hire the system. The executive’s expertise may become less scarce while the shareholder’s claim on the profits remains intact.

This was the jump in my original thought. I’d moved from intelligence becoming more accessible to economic power becoming less concentrated. Those are separate developments. One doesn’t guarantee the other.

An IMF working paper published in April 2025 makes the distinction useful. Using household data and an economic model, its authors examine how AI could narrow some wage differences while increasing wealth inequality through returns to capital. These are modelled possibilities, dependent on assumptions about adoption and how tasks change. They show why cheaper intelligence and concentrated wealth can coexist.

The question of who benefits from AI therefore reaches beyond who can use it. Who owns the infrastructure? Who controls distribution? Who has bargaining power? Who gets a claim on the gains?

A subscription gives me access to a capability. It doesn’t give me ownership of the company providing it.

Cheaper output changes what productivity tells us

I also needed to correct my assumption that anybody can now create any piece of software. Producing code is one part of delivering something useful. Security, maintenance, integration, reliability, and responsibility for failure still have to be handled.

The research gives us reasons to be specific. In a 2023 experiment involving 453 college-educated professionals, access to ChatGPT reduced the time spent on assigned writing tasks by an average of 40%, while assessed quality rose by 18%. Those results describe particular tasks under particular conditions.

A small 2025 study of experienced open-source developers found something different. Sixteen developers completed 246 tasks in familiar, mature projects. With the early-2025 AI tools tested, they took 19% longer, even though they believed the tools had helped them finish faster. That finding shouldn’t be stretched into a verdict on every developer or on later tools.

Together, the studies expose a problem with the way we talk about productivity: feeling faster, producing more, and delivering a better outcome can come apart.

Consider a team that uses AI to double the number of proposals it sends. If the proposals are poorly targeted and create more review work, the extra output may consume value. If the team serves the right clients with less administrative work, the same technology may create room for better service or shorter hours. That’s an illustrative example, but the distinction matters whenever we choose a metric.

Hours worked and units produced can help diagnose a process. They become dangerous when we ask them to explain the worth of a person.

If AI makes more output available, we’ll need to get more precise about which outcomes deserve our effort. Otherwise, we can end up working harder to supervise a growing pile of things nobody needed.

Where consciousness enters the AI economy

This is where my original question becomes less economic and more human.

There’s a lot of language circulating about consciousness determining how we experience the next stage of humanity. I can work with part of that idea, provided I’m clear about what I mean.

By consciousness here, I mean our ability to notice what’s influencing us, examine our motives, and choose a response. I’m not proposing a measurable spiritual ranking that predicts who will prosper in an AI economy.

A tool that helps me clarify a problem can also help me justify an impulse. I can ask it to examine an assumption, or I can keep rephrasing the question until it produces the answer I wanted. More capability gives me more ways to act on whatever is driving the interaction.

My hypothesis is that as execution gets cheaper, the quality of the intention behind it becomes more consequential. A poorly chosen goal can now generate a great deal of activity before anyone stops to question it.

That doesn’t mean awareness protects someone from losing their income. Rent, ownership, policy, and access remain material conditions. Explaining another person’s economic hardship as a failure of consciousness would let us avoid examining those conditions.

There are two responsibilities here: paying attention to how we use these systems, and shaping the conditions under which people have to live with them.

Could electricity become the next form of money?

I wondered whether money itself might change form, perhaps into electricity. After all, computation requires energy.

Humans are carbon-based organisms, and today’s AI runs mainly on silicon-based electronics. That shorthand is broadly accurate, but it doesn’t explain what kind of economy follows. AI is a computational system running on hardware. Its economic effects depend on what it can do, what it costs, and who controls it.

Tokens need similar care. They’re pieces of model input or output, often used for billing. They aren’t fixed units of intelligence or electricity.

Electricity could become more strategically important without becoming everyday money. As the Bank of England explains, money serves several functions: we use it to exchange, set prices, and hold purchasing power. A crucial production input doesn’t automatically serve all those functions well.

A kilowatt-hour available in one place at noon isn’t interchangeable in practice with one delivered somewhere else at night. Storage, transmission, and timing matter. We could trade claims on energy, but those claims would still need rules and trusted ways to settle them.

So I’d keep electricity-backed money in the category of speculation. We don’t need it to become reality for the underlying question to matter: what happens when access to computation becomes an important source of economic power?

What should we measure when more work can be automated?

I don’t take any of this as a reason to stop earning. Money pays for food, shelter, care, and the freedom to refuse arrangements that would otherwise control us. Dismissing it is much easier when those needs are already covered.

What I’m questioning is the habit of letting a financial measure quietly become the purpose it was supposed to support.

For someone building a business, I’d start with three questions:

1. What is enough to sustain the work? Name the income, reserves, and commitments involved. Make the requirement concrete enough to plan around.
2. What improvement are we responsible for? Choose an outcome a customer, colleague, or community can actually recognise. Revenue can help test demand; it doesn’t describe every consequence of meeting it.
3. What happens to the time AI saves? Decide whether it becomes deeper service, recovery, learning, or additional production. Leave that undecided and the old expectations can simply expand to fill the space.

These questions keep economic reality in view while giving us a way to judge what the economy is doing for us.

What deserves our time?

I don’t know whether AI will weaken capitalism, concentrate it, or produce a mixture that varies sharply between places and industries. Technology alone won’t settle that. Ownership, institutions, and collective choices will shape the result.

What I do think is becoming harder to avoid is the question underneath productivity culture. If a machine can perform more of the work I once used to prove my value, how do I decide what deserves my time?

Money can remain a practical necessity. Productivity can remain a useful measure. Neither can choose a worthwhile life on my behalf.