# On Intelligence and the Machinery of Institutions What is intelligence? The honest answer is that we do not know. But we know what universities measure, and we call it intelligence, and the machines have beaten us at it. This matters more than it should, because we built institutions around this mistake and now we cannot unmake them without admitting the mistake was always visible. Let me be plain about what happened. Universities saw the shift coming. They understood, at some level of awareness that lived in committees and strategic plans, that machines would eventually outcompute humans at retrieval, classification, pattern-matching—at the work of taking what is known and reproducing it accurately. This is the tier of intelligence we built our entire educational apparatus to measure. Tests. Exams. The recitation of facts. The correct answer from a multiple choice. These are measurable. They produce numbers. Numbers go in databases. Databases tell you whether your institution is working. So the universities did nothing. Or rather—and this is the crucial distinction—they could not afford to do anything, because the moment you try to teach judgment, you cannot measure it the same way. You cannot put wisdom in a spreadsheet. You cannot compare it across cohorts. You cannot defend your budget increases with it. This is not a failure of intelligence. It is a failure of something else entirely. --- **The Tier We Chose** There are real differences in how minds work. Some people retrieve facts quickly. Some see connections others miss. Some know when they don't know—and this last one is rarer than it should be. The tier universities chose to teach and measure was the first kind: fast, reliable retrieval. It is the tier that machines now own completely. Not because machines are intelligent. But because that particular tier—call it *recall*—was always going to be won by machines. It was only a question of when. The universities were not stupid. They were not blind. Many people inside universities understood this perfectly well. They said so. They wrote papers about it. And then they went back to grading essays on whether students remembered the dates of the French Revolution, because that is what the institution could afford to measure. I do not say this to blame individual teachers. Most teachers want to teach judgment. Most would prefer to teach it. But judgment is slow. It is local. It depends on the person being taught. You cannot scale it. You cannot standardize it. You cannot compare the judgment developed in Professor Smith's seminar with the judgment developed in Professor Jones's seminar without making a mess of the comparison, and the institution has no appetite for mess. --- **What Judgment Actually Is** Intelligence at the tier that matters—the tier machines cannot own—is not about knowing more. It is about knowing better. It is about knowing what you do not know. It is about seeing the gap between what a thing is and what people say it is, and having the nerve to say so. This is not taught. It is caught, if it is caught at all. You catch it by watching someone practice it. By doing it yourself under pressure, in front of someone who knows the difference between pretending and actually thinking. Wisdom is what you get when you have made real mistakes and had to actually live with them. Not studied mistakes. Not hypothetical mistakes. Real ones. The kind where you cannot just retake the exam. Universities cannot manufacture this. They can only create conditions where it might grow. But creating those conditions means giving up measurability. It means accepting that you cannot report to your funding bodies that you have increased wisdom by fifteen percent. It means trusting teachers and students in ways that make administrators uncomfortable. It means admitting that the tier you have been optimizing for was always the wrong tier. --- **The Cost** Here is what troubles me most: the institution can see this. It is not hidden. Any serious person who has spent time inside a university knows that something has been sacrificed for measurability. Many of them say so. And yet the machinery grinds on, producing graduates who are excellent at recall and uncertain about judgment, precisely because the machinery cannot afford to change without breaking. Who bears the cost? The students bear it. They have spent years training for a competition machines will always win. They graduate believing intelligence is something you demonstrate by knowing things, which is exactly backwards. The society bears it. We are now full of people who can retrieve information at unprecedented speeds and have almost no practice at knowing what to do with it. At deciding which questions matter. At recognizing the difference between a clever answer and a true one. The institution itself bears a cost, though it does not yet know it. The moment machines own the tier of recall completely—and that moment is now, or very near—the universities will have nothing left to teach that is worth teaching, because they have spent generations training themselves to teach only that tier. They will have optimized themselves into irrelevance. But they saw it coming. This is the thing that cannot be unsaid. They saw it, and they continued anyway, because the alternative required admitting they had built their entire structure on a mistake. --- **What Would It Take** To teach judgment in an institution designed to reward recall, you would need to do several hard things at once. You would need to measure differently, or accept that some things worth teaching cannot be measured in ways that satisfy a spreadsheet. You would need to trust teachers enough to let them teach badly sometimes, in service of teaching well. You would need to be willing to have some students fail—not because they lack intelligence, but because they lack judgment, and judgment takes time. You would need to tell the truth about what universities are for. Not credential factories. Not data points for rankings. Places where people learn to think in the presence of people who have thought hard about difficult things. You would need to accept that this cannot be scaled. It cannot be standardized. It cannot be the same in every institution. It will be slower and messier and less comparable than what we have now. And you would need to do all of this while the machines are getting better at the tier you used to own, while your funding is threatened, while parents are asking why their children are not getting ahead. This is why it will not happen. Not because it is impossible. But because the cost of admitting the mistake is higher than the cost of continuing to make it. The machinery will grind on. The students will graduate. The machines will get better. And somewhere inside the institution, someone will know that this was visible all along, and nothing was done. That is not a failure of intelligence. It is a failure of will, which is worse, because will is something we might actually control.