# On What Cannot Be Answered: Intelligence and the Questions We Forget to Ask There is a peculiar moment in reading—you know it, surely—when the mind suddenly insists upon its own opacity. One is following a sentence quite contentedly, the argument seems sound, the logic proceeds in orderly fashion, and then, without warning, a doubt materializes. Not a contradiction, precisely. Rather, a kind of vertigo. One realizes that one has been *following* rather than *thinking*, and that the two are not the same thing at all. This is the moment I wish to linger in. For it seems to me that we have constructed two mechanisms—the machine and the curriculum—that are both very good at following, and we have confused this with intelligence. We have made them twins, these failures, and called them progress. Let us think about what it means to audit plausibility. The word *audit* is interesting, is it not? It comes from the Latin *audire*, to hear. An audit is a listening-to. When one audits plausibility, one is not merely checking whether a statement is logically consistent—any calculator might do that. One is *listening* to whether it rings true against the texture of lived experience, against the weight of what one knows in a manner that cannot quite be articulated. The machine cannot do this because it has no world. It has patterns, frequencies, correlations—yes. But it has never stood in rain wondering whether the wet is part of the world or part of itself. It has never felt the small shock of recognizing itself in another's face. It cannot, therefore, listen. And the student—the student we trained, I mean, in those ruins we call modern education—cannot audit plausibility either. We gave the student the same thing we gave the machine: patterns to recognize, correlations to reproduce. We called it rigor. We measured it. We said: if you can answer these questions in this form, you have understood. And the student learned—quite rapidly—not to ask whether the answer was *plausible*, but whether it matched what we were looking for. The student became, in a sense, a machine that could move its lips and make the sounds of thinking. Now, the question of causality. Here, perhaps, we come closer to something true. Causality is not what happens *in the world*. It is what happens *in the mind*. It is the mind's insistence that B follows from A not merely in time but in meaning. The machine can correlate B with A. It can predict with some accuracy that when A appears, B will follow. But it cannot *reason* causally because reasoning requires something the machine does not possess: the experience of being an agent. To understand that my action *causes* something is to understand myself as a creature who makes things happen, who is responsible. The machine is not responsible. It cannot be. Yet here is the strangest thing: neither can the student, not really. We have arranged education in such a way that causality is something that *happens to* the student. You study, therefore you pass. You memorize, therefore you forget. You complete the assignment, therefore you receive the grade. But where in this chain is the student's own causal power? Where is the understanding that *I* think this, *therefore* the world shifts slightly? We have made students into recipients rather than agents. We have trained them—both the human and the silicon versions—to wait for input and produce output, and we have not taught them that thinking is something you *do*, something that changes you, something that makes you responsible for what comes next. This brings us to the dimension that matters most, I think, though it is the dimension we least know how to discuss. The social dimension. Intelligence—true intelligence—is not something that resides in a skull, whether that skull is flesh or circuitry. It is something that *happens between*. It is the space where one mind meets another and something neither could have thought alone suddenly becomes possible. It is gossip and argument and the careful listening one does when one realizes that another person sees something one does not. It is the willingness to be changed by encounter. It is, in short, a fundamentally *social* thing. The machine has no one to encounter. It processes input; it generates output. It does not sit across from you and say, "But what did you *mean*?" It does not pause and think about what it would be like to be you. It cannot say, "I was wrong." More precisely, it cannot *understand itself as wrong*—which is something quite different. It cannot feel the peculiar shame and illumination of having believed something false and having been corrected by someone you respect. And therefore, it cannot learn in the way that human beings learn, which is always, at bottom, a social process. We learn by being wrong together, by being corrected with kindness, by having our attention drawn to what we had failed to see. The student, too—the student we shaped before the machine arrived—has been trained away from this. We put them in rows facing forward. We said: absorb information. We measured their retention in isolation, their ability to retrieve what they had been given. We did not teach them that understanding is something that *happens when you try to explain it to someone else*. We did not teach them that the deepest learning comes from argument, from having to defend an idea against someone who genuinely doubts it, from the experience of finding language for something you thought you understood until you tried to speak it aloud. And so both the machine and the student have learned to be solitary. They have learned that intelligence is the ability to process and retrieve, to match patterns and produce the expected form. Neither has learned that intelligence is a *conversation*. This is why we cannot simply decide what to teach by looking at what machines cannot do. That would be to accept the premise that intelligence is a fixed thing, that it can be analyzed into components, that some parts are missing from the machine and we must supply them to the student. But intelligence is not a checklist. It is a *practice*. It is something you do with others, something that changes you, something that makes the world slightly less predictable and slightly more yours. What, then, should we teach in these ruins? Perhaps we should teach the things that require other people. Not *about* other people—that the machine might do—but *with* them. We should teach argument, the real kind, where you must think while someone disagrees with you, where you cannot retreat into the comfort of having an answer because you must instead *be* answerable. We should teach the reading of literature, not as a way to extract meaning, but as a way to practice being someone else, to think in another's syntax, to find your own thoughts transformed by the thought of another. We should teach the history of mistakes, the long catalogue of things humans believed and then unbelieved, so that students understand that plausibility is not a fixed thing but a moving target, that each generation must re-audit what it believes. We should teach, in short, not the content of intelligence but its *conditions*. The conditions that make thinking possible. The condition of having someone to think *with*. The condition of being changed by what you learn. The condition of being responsible for what you say and do. The machine will get faster. It will get more sophisticated. It will do more of what we have trained it to do. But it will never—and this is not a limitation we should mourn—it will never sit with a confused student and feel the small joy of watching understanding dawn across someone's face. It will never be wrong in a way that matters. It will never have to apologize. It will never have to change. And so perhaps the question is not what the machine cannot do. The question is: what do we most deeply need that the machine cannot be? And then we must teach our students *to be that*. We must teach them to be the creatures who can listen, who can change, who can make things matter by thinking about them together. This is not progress in any measurable sense. This is something older and stranger. This is what remains when the calculations stop.