AI Is Removing the Administrative Weight of Civilisation From the Human Mind

If we go back through our evolution, we were hunter-gatherers, then agriculture, then modern life. All along we were curious, and our survival depended on coordination. So we coordinated to survive. That is our cooperation.

Then, to cooperate without losing each other in the noise, we made language. For the mobility of knowledge, we found writing. To convey the nature of the world and keep working on it, we found numbers and mathematical notation. Then books. Then the printing press. Then the internet.

Every one of those was the same move. Take something distributed, scattered, sitting inefficiently in nature and in our heads, and compress it into one place where it can be stored, retrieved, and used by everyone.

That is how we evolve: compression. Progress is not some moral arc. Progress is finding efficiency through compression systems.

Now AI is compressing the rest of it. The judgment-work, the decision-work, the administrative weight we were carrying in our own small memory because there was nowhere else to put it. The same footsteps, the same direction.

And we call it doomsday.

We are surprised by the most unsurprising thing our species has ever done.

The short version:

• Progress is not a moral arc. It is a ten-thousand-year compression sequence: cooperation, language, numbers, writing, print, internet, AI. Each layer removes inefficiency by pulling what the mind was holding into a place outside the mind.

• AI is the next layer in that sequence. It is not breaking the line of human development. It is the line, arriving on schedule.

• What AI takes is the administrative weight of civilisation: the storage, the retrieval, the routine decisions the mind was forced to carry because there was nowhere else to put them.

• The risk is real but narrow. One 2025 study of 666 people found that heavy AI users scored measurably lower on critical thinking, and the thing in the middle was cognitive offloading. That is the boundary condition, not the headline.

• What is left when the weight comes off is the only thing that was ever yours: your expression, and the judgment that directs it. At Ivanooo we call that the Judgment Layer.

The move we have been making for ten thousand years

Ian Morris, the Stanford historian, built an index to measure civilisations across fifteen thousand years. He did not measure them by their art or their kings. He measured four things: energy capture per person, the size of their largest cities, their war-making capacity, and their information technology, meaning the speed and reach of writing, printing and telecommunication.

Look at that list again. Two of the four are compression.

Energy capture is taking diffuse energy, sun and muscle and wood and coal and oil, and concentrating it so one person can do the work of fifty. Information technology is taking diffuse knowledge and concentrating it so one person can reach what a thousand people learned. The measure of a civilisation, in large part, is how much inefficiency it has squeezed out of survival.

So when we made language, we compressed meaning into sound. When we made writing, we compressed memory into marks that outlived the one who remembered. When we made mathematical notation, we compressed the structure of the world into symbols we could keep working on without holding the whole thing in one head. Each step took our collective knowledge, made it less random, less scattered, and made acting on it more efficient.

This is the part people skip. They treat AI as a rupture, a thing that happened to us. It is not a rupture. It is the most recent entry in a list our species has been writing since before we had a word for it. I went deeper into this in The Efficiency Theory of Human Evolution. The footstep does not change. Only the layer being compressed changes.

For all of that history, we compressed information outside the mind. AI is the first layer that compresses the work happening inside it.

What AI is actually compressing

Here is the thing we collected so carefully and never noticed we were carrying.

For most of human history, knowledge sat distributed. In nature, in other people's heads, in scattered books, in your own overloaded memory. Our obsession with inefficiency made us pull all of it into one place so it could be retrieved and used by anyone. Libraries did this. The internet did it faster. AI does it again, and this time it does not just store the knowledge. It operates on it.

What it lifts off you is administrative weight. The looking-up. The remembering-where. The routine synthesis. The generation of ridiculous work, the thousandth version of a document that never needed a thousand-and-first original thought.

The mind already wanted this relief. In 2011, the psychologists Betsy Sparrow, Jenny Liu and Daniel Wegner ran four studies and found something quiet and enormous. When people expect to have future access to information, they remember the information itself less, and they remember where to find it more. The internet had already become external memory. A partner the brain treats like a person who knows things, so you do not have to.

We have a limited amount of memory. Our memory is storage space. We were burdened with a surplus of information. So our quest to find a solution arrived. AI is that same instinct, finishing the move.

So, it is egalitarian, right?

That is the claim people want to make, and the one I would be careful with. The knowledge gap, the thing credentials and gatekeepers were built to guard, genuinely compresses. Access opens wider than any layer before it. But access is not capability. A library open to everyone did not make everyone a scholar. When the knowledge becomes free, the thing that separates people stops being what you can retrieve. It becomes what you can do with it. Judgment, and whether the systems that route opportunity can find you at all. The compression makes the input cheap. It makes the output expensive.

Then why is it doomsday?

Then why are we against it? Why are we surprised? Why does it feel like the end?

Because we are relating it to the wrong thing. Look closely and the fear is not about productivity, not even about the job. It is identity. For two centuries we told people that what they produce is who they are. So the job became the self. Now a machine produces the output, and it does not feel like a tool arriving. It feels like the self getting deleted.

That is not AI taking your mind. That is AI showing you how much of your identity you had quietly rented to your administrative function. I wrote the full version of this in Who Are You? The Question AI Is Forcing All of Us to Answer Again and in The Construct You Did Not Choose. The short form: the layer being shaken is not the human layer. It is the construct we mistook for the human layer.

So see this clearly, and do not relate it to judgment. You do not need to do that. See it as the compression of a long-pending burden. The routine decisions, the symbolic load, the distributed knowledge that was always supposed to live in external memory, and was sitting in your head only because external memory had not caught up yet.

The doom is real as a feeling. It is wrong as a diagnosis.

The line you must not cross: do not offload the judgment

Now the part I will not soften, because this is where the reframe earns its keep.

There is a version of this that goes badly, and the research is already showing its shape. In 2025, Michael Gerlich studied 666 people across age groups and found that heavy AI use lined up with measurably lower critical thinking. The thing sitting in the middle was cognitive offloading, which he described as a falling away of the inclination to think deeply at all. An MIT Media Lab study pointed the same way, towards what they called cognitive atrophy. A study of consultants found AI lifted short-term output by 14 to 40 percent while quietly eroding the independent judgment it was meant to support. Metrics up, capability down, and you find out too late.

So here is the line. Compressing the work is the oldest human move, and it is good. Compressing the judgment is not compression at all. It is what we call cognitive surrender, and it is the one thing in the whole sequence that does not externalise cleanly.

Language did not make us worse at meaning. Writing did not make us worse at thinking. Each of those offloaded storage and transmission, the Layer 1 work. AI is the first layer that can quietly absorb Layer 2, the deciding, the evaluating, the catching of your own error, if you let it.

That is the hidden trade nobody prices in. The administrative weight comes off whether you are paying attention or not. The judgment only stays if you keep using it.

What AI takes versus what stays yours

Layer	What it covers	What AI does to it	What it costs you if you stop using it
Layer 1: Efficiency	Storage, retrieval, calculation, routine synthesis, first drafts	Compresses it, far better than you can	Nothing. This was administrative weight. Let it go.
Layer 2: Judgment	Deciding which problem to solve, judging whether the output fits your context, catching the error	Absorbs it if you hand it over, but cannot actually carry it	Everything. This was never administrative. This was you.

The four operators that Ivanooo measures, Generating Alternatives, Revising Beliefs, Connecting Patterns and Tracing Consequences, are what Layer 2 is made of. They are the substrate of judgment, and like any capability they strengthen through use and waste away through disuse. Machines took the body's load and we built gyms for it; AI is taking the mind's load, and the same logic holds, which I wrote about in Machines Took the Body's Load. We Built Gyms. AI did not change that biology. It only removed the friction that used to force you to exercise it.

This is the whole game now. Hand AI the weight. Keep the judgment. The people who confuse the two will take the productivity bump and lose the very thing the bump was supposed to free them for.

What is left when the weight comes off

Take away the administrative function, and a question is left standing that most people were always too busy to ask. If you are not the work, then who are you?

The answer the species kept pointing at, across every philosophy and every life that felt fully lived, is the same one. Life is the expression of ourselves into the world. Expression through experience. That is how we evolved. That is how the brain works. That part was never going to be compressed, because it is not inefficiency. It is the point.

All that we created was not for becoming a small-time goon and identifying with that identity. No. We are identified with our expression. The looking-up and the remembering-where and the generation of ridiculous work, that was never the expression. It was the toll we paid to reach it.

AI is humanity trying, the way it has always tried, to remove the administrative weight of civilisation from the human mind. So the mind can finally spend itself on the thing it was always for.

The weight is coming off. What you do with the lightness is the only question left. And that is judgment. That is yours.

---

FAQ

Is AI making us dumber?

Not by itself. A 2025 study of 666 people found heavy AI users scored lower on critical thinking, but the cause was cognitive offloading, not the tool. AI compresses the routine work, which is good. The damage comes only when you also offload the judgment: the deciding, the evaluating, the catching of your own error. Tools do not dull the mind. Lazy use of them does.

What is cognitive offloading?

Cognitive offloading is using something outside your head, a notebook, Google, AI, to do mental work you would otherwise do inside it. It is ancient and mostly healthy. Writing offloaded memory. Calculators offloaded arithmetic. The risk is narrow. Offloading storage and retrieval frees you. Offloading judgment and reasoning wastes the very capabilities that made you valuable.

Does using AI weaken critical thinking?

It can, but it is not automatic. Research from Gerlich in 2025 and from the MIT Media Lab links heavy, passive AI use to weaker reasoning and cognitive atrophy. What decides it is how you use it. People who treat AI output as a finished answer decline. People who treat it as a first draft to interrogate keep their judgment sharp and gain the speed on top.

Is AI access actually equal for everyone?

Access is wider than any earlier layer of knowledge technology, but access is not capability. A free library did not make everyone a scholar. AI compresses the knowledge gap that credentials were built to guard, which is real and significant. But it raises the value of what you do with that knowledge, judgment and visibility, and those are not handed out equally just because the input now is.

What makes us human in the age of AI?

The part that was never administrative: judgment and expression. AI can compress storage, retrieval and routine synthesis, the load civilisation forced onto the mind because there was nowhere else to put it. It cannot carry the deciding of which problem matters, the judgment of whether an answer fits your life, or the expression of one specific consciousness into the world. That part is the point, not the overhead.

Why does AI feel like an existential threat if it is just another tool?

Because for two centuries we taught people that what they produce is who they are. When a machine produces the output, it feels like the self being erased rather than a tool arriving. The threat is to a borrowed identity, the construct that mistook the job for the person. Not to the human underneath it.