AI and society / 26 Jul 2026

The blade of grass theory

Hand a household robot a small pair of scissors. Tell it to cut the lawn to one centimetre, blade by blade.

It will do it.

Not because it is clever. Because eight hours crouched in the sun costs it nothing. No boredom, no sore back, no wage, no quiet resentment about who got given the job.

That is the part I keep turning over. Not what robots will be smart enough to do. What they will be willing to do, for how long, at what price.

The headline was the least interesting part

Musk told The Economist last week that money will not matter in 2036. It travelled as a prophecy, which is the least useful way to read it.

The argument underneath the headline is a supply argument.

If AI passes the sum of human intelligence in roughly five years, and that intelligence gets poured into machines that can move, then goods and services start arriving faster than people can consume them. Supply running that far ahead of demand pushes prices down, not up. So he expects deflation rather than inflation, and governments writing cheques rather than employers paying wages. He adds the obvious caveat, which is that none of this happens if we blow ourselves up first.

You do not have to accept the date to find the mechanism worth understanding.

Patience is the input nobody prices

The constraint on physical work has almost never been knowledge.

Farmers know how to weed. Builders know how to build. Nurses know what care looks like. The constraint is that human attention is finite, expensive, and hates repetition.

So we designed around it. We spray the whole field because inspecting every plant is unthinkable. We build to a schedule set by how long people can safely work. We accept that a house takes months and a hospital takes years.

Every one of those decisions is a workaround for the same shortage.

Now picture that shortage easing. Not intelligence getting cheaper, which has already happened. Physical patience getting cheaper. Attention that can be pointed at a dull, repeated, bodily task and simply left there.

Farming. Weeding plant by plant instead of spraying by hectare. Picking each piece of fruit at its own ripeness rather than the average ripeness of the block.
Building. Sites that run through the night because nothing on them gets tired. Materials placed to the millimetre because precision no longer costs extra hours.
Care. Someone in the room at three in the morning, every night, without burning out a daughter who also has a job.
Cleaning and maintenance. Upkeep as a continuous background process rather than an event you schedule and dread.
Mining, logistics, manufacturing. The dull and the dangerous, which is most of what we currently pay danger money for.

Most prices are hours in disguise

Almost everything expensive is expensive because of the human hours inside it.

A house is timber, bricks and about eighteen months of skilled attention. Care work is unaffordable to buy and underpaid to do at the same time, because it is nearly pure hours. Even the things that feel like materials are mostly somebody's shift.

Take patience out of the cost of a thing and the price structure underneath it changes shape. That is the whole deflation argument, and it is a more serious claim than the headline it got attached to.

What robots cannot make

Robots make things. They do not make everything.

Land in a particular place. A beachfront block does not multiply because manufacturing got good. Anything scarce by position stays scarce.
Energy and materials. Robot labour is not free labour. It is energy converted into work, at scale, continuously. The constraint moves from human hours to electricity, minerals and compute, which is a real constraint with real politics attached.
Anything positional. Status is defined by other people not having it. Abundance cannot fix that by definition.
Trust, judgement, and one specific person's attention. Still scarce. Possibly more scarce.

So money does not vanish. It concentrates onto the shrinking set of things that cannot be manufactured. That is a different world from the one where money stopped mattering, and it is the more likely one.

Abundance does not distribute itself

There are two claims tangled together here, and only one of them is about technology.

Cheap production is an engineering claim. You can argue about the timeline and still see the trend.

Everyone being better off is a political claim. Nothing inside the robot decides who owns it. Deflation is a gift if you have income and a threat if your income was the thing that got automated. Universal high income is a policy that would have to be chosen, funded and defended, repeatedly, by people who disagree.

The machines arriving is the easy part. It is the second claim that is genuinely uncertain, and it will not be settled by better hardware.

Prompt you can use

I want to stress-test one assumption about a business.

Here is what it does: [describe the business in a few lines].

Identify the parts of its cost base that are mostly human hours spent on repetitive physical work or repetitive attention. For each one, tell me:

1. What that work costs today, roughly, and why it costs that.
2. What the job would look like if that patience cost close to nothing and could run continuously.
3. What would still be expensive afterwards, and why.

Be specific to this business. Do not write a general essay about automation.

The rule

Do not ask what robots will be smart enough to do.

Ask what becomes possible when patience stops costing anything, then ask what is left that no amount of patience can produce. That second list is where the next few decades get decided.