The Friction We Keep Removing Is What Makes Us Human

The Friction We Keep Removing Is What Makes Us Human

A warehouse in Beijing swapped its night shift for machines, not because the robots became clever enough but because a cost finally fell below a wage. That is a moral event dressed as arithmetic, and almost nobody paused over it.

Let me say where I stand before the evidence, because the evidence will tempt you toward the wrong conclusion. We are rapidly moving to an era where intelligence is becoming practically free, and that will reshape society in ways we cannot yet picture. That is why I also do not believe we are in a bubble. Railways and telecom were financed on capital that looked reckless at the time, and we ended up needing everything they built and more. Anything that can be built will be built, and most of it should be.

So this is not a plea to stop. It is a harder question. What are we building, and does anyone actually understand it?

Which demand can you actually touch?

Take the demand question first. China shipped more than 40,000 humanoid robots in six months because the unit economics closed, while Nvidia guaranteed up to $105 billion so its own customer could build the data center that will run its chips. Both are answers. Only one is demand you can touch. Which kind sits underneath your own AI investments?

That distinction matters because the loud story was the reversible one. A financing guarantee can be revised, and this one already was, by $145 billion. The quiet stories were the ones you cannot take back.

Four arms of the US government defined what a digital dollar is: what it legally counts as, who may issue it, whether a treasurer may hold it as cash. Because the rule reaches any dollar token issued to a person on American soil, a domestic definition became a global default. A bank in Singapore inherited rules it never voted on.

A national health service opened a standing framework to buy mixed reality as routine clinical training, while the companies that sold the consumer metaverse quietly dismantled it. A procurement lot settled in one line what a decade of platform wars could not. And medicine arrived for the single patient: a cancer vaccine built from one person's own tumor, a brain operation guided by live software, an Alzheimer's blood test cleared for use. None of it framed as an experiment. All of it billed as care.

Not one of these was decided by an invention. Every capability existed earlier. What changed is that an institution committed, and at that moment the question stopped being a question. It became a definition. A framework. A clearance. A purchase order.

The people who were not asked

Who was in the room when those were signed? Not the nurse who will train on that system. Not the worker whose shift was costed against a machine. Not the patient whose scan a model reads before a doctor does. The people who inherit a decision are almost never the people who make it, and we have built a way of deciding that makes this faster, not slower.

Humanity does not think in centuries. Had we held a long view a hundred years ago, we would likely be somewhere better than here. We are pouring the foundations of the next century, silicon first, then photonics, perhaps one day organic again if we learn to match the efficiency of the human brain, and we have not agreed what the century is for.

My answer to my question at the start would be to build it to elevate human consciousness, awareness and empathy, in a way the planet would also endorse. That is a choice, but it seems most of our current decisions are made by default, at the speed of a purchase order, without taking a long-term perspective.

Each decision removed a friction

Look again at what those four decisions share. Each one removed a friction. The friction of negotiating with a worker. The friction of legal ambiguity. The friction of deciding what to buy. The friction of a diagnosis that once needed a spinal tap and a specialist few could reach.

But there is a cost that nobody prices. A robot is cheaper than a wage until the factory that built it goes bankrupt and you are left with a worker that no longer works.

When solar crossed below oil and gas, the crossing was a beginning, not a verdict. The same crossing has arrived for human labor, but in this case the hidden costs are still missing from the equation. The move to renewables carries exactly this kind of unpriced cost, a mountain of discarded wind turbine blades that nobody budgeted for. How many of our efficiency gains carry a liability nobody has written down?

The largest of those unpriced costs is friction itself.

Machines optimize, and optimization means removing friction. They will learn to perform awe, joy and care convincingly, and many people will find the performance sufficient. Some already prefer it because there is no friction, as these tools are more yes-sayers than honest companions.

There is a Dutch saying, "wrijving geeft glans". Friction gives shine, if I were to translate it literally. Sand away the friction of people being difficult, mistaken and human, and you get a smoother society and a duller one, having surrendered far more than the transition returns.

So I hold two positions that pull against each other. Build all of it, but let us slow down, because we do not understand what we are building. The way through is not a compromise between them. It is sorting. There are plenty of optimizations in society that carry real benefits with few downsides, that can make our society better in the short run without damaging consequences in the long run. However, there are also plenty that require more time to decide.

Where the work begins

Because of the speed of change, the world is changing faster than we can absorb. But if we want to build for the long term, absorption has to become a practice rather than a private virtue.

That has to start early, long before anyone reaches a board. We teach children how to use these tools. We do not teach them to ask what a tool costs ten years after it works. Without a deep understanding of what is at stake, we are bound to make mistakes we cannot take back.

So let us decide first which decisions carry trivial consequences and which carry consequences measured in decades. Then move fast where we can and slowly where we should. Most organizations do the reverse, deliberating for months over choices they could undo in a week and settling permanent ones before lunch.

The machines will keep removing friction. That is what they are built to do. Our work is to decide which friction was holding the shine, and to defend it before it has been polished away. If you cannot yet say which one, that question is where the work begins.