What does IQ actually measure?

IQ measures administrative intelligence: a cheap, fast, statistically stable proxy built to sort people at scale when societies grew too large to observe everyone in context. It correlates decently with performance on symbolic, abstract tasks, which were scarce and valuable in the early knowledge economy. It was never a measurement of something fundamental about humans. It was legibility, not truth. The error came when we forgot it was a proxy and started saying intelligence is what IQ measures.

Why is AI making IQ obsolete?

AI can do the things IQ measures: symbolic reasoning, information retrieval, abstract problem-solving, pattern matching across large datasets. It scores 90th percentile on the tests built to measure human intelligence. That forces a choice. Either AI is intelligent, or those tests never measured what we thought. They measured the bottleneck of a previous technological era. The cognitive labor that was scarce in 1950, locked in human heads, is now commodity. The proxy lost its economic foundation.

Is IQ real, or was it always a construct?

IQ is real in that it is cheap to administer, statistically stable, and correlates with symbolic task performance. But it was built to solve administration, not to define intelligence. When societies were small, capability was obvious: you watched someone for a week and you knew. Scale broke that. You cannot observe millions in context, so you build a proxy. IQ, grades, credentials. The construct worked until we mistook the proxy for the thing itself.

What replaces IQ as a measure of capability?

Capability in context, not cognitive horsepower in isolation. Intelligence is the ability of a human plus their tools plus their environment to stay coherent, adaptive, and generative as conditions change. That means adaptation under uncertainty, judgment when there is no instruction manual, and the ability to work with tools and people across time. We now have infrastructure for continuous observation and behavioral assessment over time, so scale no longer forces reduction to a single test score.

Why did the proxy problem warp education and hiring?

Once intelligence became defined as what IQ measures, the whole system bent around it. Kids get labeled at age seven on a test score. Educational philosophies optimize for standardized measurement. Hiring filters on credentials that filter on test scores. Each layer treats the proxy as ground truth instead of correlation. We forgot it was ever administrative. The article frames this as a 100-year error now being corrected, not as rebellion but as paying attention.

What is the Growth Velocity Index?

The Growth Velocity Index (GVI) is a diagnostic tool that measures how effectively you leverage AI for professional growth. It identifies your AI interaction profile and provides actionable insights.

How long does the GVI assessment take?

The GVI assessment takes approximately 3-5 minutes to complete and provides immediate results with personalized recommendations.

What are the Four Operators in the GVI framework?

The Four Operators are classification profiles for AI interaction patterns: Operator 1.0 (Traditional), Operator 2.0 (Experimenter), Operator 3.0 (Integrator), and Operator 4.0 (Transformer).

Administrative Intelligence: The 100-Year Proxy Problem

IQ was never meant to measure intelligence—it was built to solve an administrative problem. The proxy became the ontology.

6 min read February 3, 2026

Here's what pisses me off about how we talk about intelligence.

We act like IQ is this sacred thing. This real measurement of something fundamental about humans. And anyone who questions it is either anti-science or coping.

But that's not what happened. That's not how IQ came to exist.

The original problem was never "what is intelligence." The original problem was administration.

Think about it. When societies were small, you didn't measure intelligence. You just... saw it. Could this person hunt? Could they build? When things went sideways, could they adapt? Could they coordinate with others without everything falling apart?

Capability was obvious. You watched someone for a week and you knew.

Then scale happened. Factories. Armies. Mass education. Nation-states trying to process millions of people through standardized pipelines.

And suddenly you hit a wall: you cannot observe everyone in context.

So what do you do?

You build a proxy. Something cheap. Something fast. Something that correlates well enough with the thing you actually care about.

That's IQ. That's grades. That's credentials.

Not truth. Legibility.

And look—the proxy worked. I'm not saying it was useless.

IQ is cheap to administer. Statistically stable. Correlates decently with performance on symbolic, abstract tasks. Fits bureaucratic systems perfectly.

In the early knowledge economy, this made sense. Symbolic reasoning was actually scarce. Computation happened in human heads. Information was locked up in libraries and institutions.

So measuring raw cognitive horsepower? Rational choice.

But here's where it went wrong.

The proxy became the ontology.

We started saying "intelligence IS what IQ measures." Not "IQ correlates with some aspects of capability." No—IQ IS intelligence. The map became the territory.

And once that happened, the whole system warped around it.

Kids labeled at age 7 based on a test score. Entire educational philosophies built around optimizing for standardized measurement. Hiring systems that filter on credentials that filter on test scores that filter on... what exactly?

We forgot it was ever a proxy. We started treating it as ground truth.

That's the part that gets me. We didn't just use administrative intelligence. We forgot it was administrative. We thought we'd found something real.

Now here's the rupture.

AI shows up. And AI can do all the things IQ measures.

Symbolic reasoning? Check. Information retrieval? Check. Abstract problem-solving? Check. Pattern matching across large datasets? Check.

AI scores 90th percentile on the tests we built to measure human intelligence.

So either AI is intelligent—in which case, what's special about humans?

Or—and this is the one nobody wants to say—those tests were never measuring what we thought they were measuring.

They were measuring the bottleneck of a previous technological era.

The stuff that was scarce in 1950 is now abundant. The cognitive labor that was valuable when it lived exclusively in human heads is now commodity.

This is what I mean when I say the proxy just lost its economic foundation.

For 100 years, administrative intelligence worked because the things it measured were genuinely scarce and valuable. Not anymore.

So what collapses?

IQ as the dominant signal. g-factor as the center of capability. Test scores as the gateway. Credentials as proof. The entire brain-only, individual-only, static-measurement model of human worth.

All of it was built on a proxy. And the proxy just became worthless.

But here's what's interesting. And I don't think people have caught up to this yet.

We're not losing something. We're being released from something.

For the first time in maybe ever, scale doesn't force reduction. We have the infrastructure for continuous observation. For behavioral assessment over time. For understanding capability in context rather than in a testing room.

We can go back to what intelligence actually was before we had to compress it for administrative convenience.

Adaptation under uncertainty. Coherence when everything changes. Judgment when there's no instruction manual. The ability to work with tools, with other people, across time horizons.

Not cognitive horsepower in isolation. Capability in context.

I keep coming back to this frame:

Intelligence is not internal capacity measured in a vacuum.

It's the ability of a human-plus-their-tools-plus-their-environment to remain coherent, adaptive, and generative as conditions change.

That's a mouthful. But it's closer to reality than "your brain on a standardized test."

The brain matters. Obviously. But so does the body. The environment. The tools available. Other humans. Time.

Intelligence isn't stored in your head. It's distributed across the whole system.

Now—I'm not naive about this.

The shift I'm describing is hard. Harder to measure cleanly. Threatens a lot of entrenched institutions. Requires thinking in trajectories rather than snapshots. Breaks the simple rankings that make administration easy.

Some people prefer simple rankings. They're going to hate this.

But that's not an argument for keeping a broken system. That's just inertia dressed up as caution.

Here's the historical arc as I see it:

Evolution gave us intelligence as adaptation. That's what it was for millions of years. Can you respond to what's actually happening? Can you learn? Can you adjust?

Then industrialization needed to sort people at scale. So we invented administrative intelligence. A proxy. Useful, but a proxy.

The knowledge economy made the proxy dominant. IQ and credentials became the gatekeepers for everything.

Now AI is making the proxy collapse. The things we measured are commodities. The things we ignored are the only things left.

The question isn't whether the old system is dying. It is.

The question is what we build next.

And I think what we build next looks a lot more like what intelligence was before we had to compress it. Situated. Contextual. Dynamic. Observable through behavior over time rather than performance in artificial conditions.

We're not dumbing things down. We're correcting a 100-year error.

That's not rebellion. That's just paying attention.

Short Version:

IQ was never a measure of human worth. It was a proxy built to sort people at scale.
The error: we let the proxy become the ontology. Map replaced territory. IQ became intelligence itself.
AI scores 90th percentile on those tests, which means they measured a previous era's bottleneck, not intelligence.
Real intelligence is capability in context: adaptation under uncertainty, observed through behavior over time.