# What Intelligence Is Not, and Why That Matters Let me begin with something I have observed. When a man claims to possess intelligence, he usually means one of two things: he can answer questions quickly, or he can answer questions at all. The machine does the first. The student, badly taught, does the second. Neither knows what he is doing. This is worth examining plainly. A machine trained on text can produce plausible sentences. It does this by finding patterns in what came before. It is very good at this. But plausibility is not truth. A machine cannot tell the difference. It has no way to know whether the pattern it found reflects something real in the world or merely something common in the text it read. A student taught to memorize and regurgitate has the same problem, though he usually senses it. He feels the hollowness. The machine does not. The machine cannot ask whether its answer makes sense causally. It cannot think: "If this were true, then that would follow, and I have never observed that." It cannot hold two ideas in mind and watch them collide. It cannot say, "This is wrong because I have seen the world." Neither can the student trained in the old way—the one who learned facts without learning how facts connect to other facts, how the world actually works. This is called progress by people who do not think carefully about words. Here is what intelligence actually is: it is the ability to see what is true. Not quickly. Not impressively. But truly. This requires several things that neither machine nor badly-educated student possesses. First, it requires contact with reality. You must have done something. You must have failed at it. You must have noticed what went wrong and why. The machine has done nothing. It has only read about doing things. The student, if he has been taught rightly, has at least attempted something—grown something, built something, broken something, and watched the consequences unfold. Second, it requires judgment about causation. You must be able to think: "This happened because of that, not because of something else." The machine cannot do this. It can find correlation. It cannot distinguish between correlation and cause, and it does not know it cannot. The student can learn this, but not from a textbook alone. He learns it by noticing that when he plants seeds in winter, they do not grow, but when he plants them in spring, they do. He learns it by consequence. Third—and this is where wisdom enters, which is not the same as intelligence—you must know which questions are worth asking. This is the hardest thing. A machine will answer any question with equal confidence. A badly-educated student will do the same. But a man who has lived, who has worked, who has thought about what he has seen, begins to know that some questions matter and others do not. He knows that some answers, though they sound elegant, explain nothing about how life actually goes. This is wisdom: the knowledge of what is worth knowing. I have met clever men who could answer ten thousand questions and yet could not see what was directly in front of them. They had intelligence of a sort—the ability to manipulate symbols, to remember, to construct arguments. But they lacked wisdom. They did not know what was true because they had not lived enough to recognize truth when they saw it. They had not failed enough. They had not watched the world refuse to cooperate with their theories. Now, what should we teach? Not more facts. The machine will always beat us at facts. Not techniques for passing tests. These teach a student to mistake plausibility for truth, which is exactly what we have done. We should teach contact with reality. Give the student something to build or grow or make. Let him fail. Let him notice his failure. Let him try again. This teaches causation. It teaches what matters. We should teach reading—real reading, not skimming. The student should read until he understands not just what a man said, but whether what he said is true, and how you would know if it were false. This teaches judgment. We should teach argument. Not debate, which is performance. Real argument, where two people try to discover what is true together, where they say, "You are wrong because..." and the other person must answer truly. This teaches the difference between winning and knowing. We should teach silence. The student should spend time doing nothing, thinking about what he has seen. This is where wisdom grows. You cannot hurry it. The machine will not do these things. It cannot. It has no body, no failures, no stake in truth. This is not its fault. It is simply not intelligent in any way that matters. The student who was trained in the ruins of our coincidence—trained to be fast and plausible like the machine—will not do these things either, unless we teach him that they are worth doing. He has been taught to mistake speed for understanding. He has been taught that an answer is good if it sounds right. We must teach him differently. We must teach him to be slow. To be suspicious. To ask: "Is this true? How would I know? Why does it matter?" These are not the questions a machine asks. They are the questions a man asks when he is trying to live rightly in a world that does not explain itself. This is what intelligence is. Not brightness. Not quickness. The ability to see what is true, and to care whether you see it truly.