# On Swiftness and the Obscuring of Mind There is a peculiar moment—one recognizes it by a certain quality of suspended breath—when one realizes that something has been decided without quite deciding. The machine generates. One reads. The reading itself feels like assent. And here we arrive at a question that cannot be asked in the register of blame, though blame will certainly follow, but rather in the register of *becoming*: what happens to accountability when speed enters the room? Let us begin where intelligence research does not often begin: with a cup of tea growing cold. With the moment one looks up from a page and thinks, *I have just read something, but did I read it?* The machine's fluency has a quality one might call hypnotic. It speaks without the hesitations that ordinarily punctuate human utterance—those small silences in which doubt might nest. It does not say, "Perhaps," or "One wonders if," or trail into the murmur of qualification. It arrives at the appearance of finished thought. And here, I think, is where we must pause. Intelligence, as it has traditionally been measured and admired, has always contained within it a certain human *friction*. The scholar who rewrites a sentence four times is engaging in something more than mere refinement; she is thinking *against* the resistance of language itself. The engineer who sketches and crosses out is not wasting time—he is allowing doubt to generate new pathways. But the machine encounters no such friction. Its speed is not the speed of genius breaking through obstacles; it is the speed of paths already laid. And yet—and this is the difficulty—we have been taught to read speed as a marker of excellence. Someone decided, did they not, that this speed was desirable? This is where the social dimension enters, quietly, the way important things often do. It enters not as an afterthought but as the very architecture of the problem. For the question "Who decided that speed was acceptable?" contains within it a smaller, preceding question: *In whose interest does speed operate?* And this question opens outward like a door one didn't notice was there. Consider the texture of modern institutional life. There is a kind of velocity built into its very bones—not because velocity is inherently valuable, but because velocity *reduces friction*, and friction is where questions live. If one can move quickly enough, one need not answer the slow questions. One need not sit with a thing long enough for its wrongness to become apparent. The machine, in this sense, is not an anomaly in the landscape of contemporary decision-making; it is rather an intensification of a logic already present. It is what happens when the desire to eliminate friction meets the technology to do so. And then three weeks pass. The error surfaces. It surfaces not in the reading but in the living—in the world where things matter, where consequences accumulate, where people are affected by decisions they did not know were being made at such velocity. Here is where accountability becomes nearly ungraspable. For accountability ordinarily assumes a *person*, someone who decided, someone whose judgment was in question. But what we have instead is a distributed event. The machine generated. You read. The reading took perhaps ninety seconds. The machine's operation took less than that. The decision to deploy the machine, to trust its fluency, to accept its output without the time for slowness—*that* decision has many authors, and yet none of them seems quite to be the one holding the pen when the moment comes to assign blame. This is not a failure of accountability *mechanisms*, though it is sometimes discussed as such. It is rather a problem of what one might call *perceptual velocity*—a mismatch between the speed at which something happens and the speed at which human attention can track it. We are asked to stand in a current moving faster than our capacity to feel its motion. And in that condition, the very notion of accountability becomes slippery, difficult to grasp. But let us not retreat into the comfortable abstraction that "the system is responsible." For that too is a way of dissolving the question. Someone, somewhere, made a choice to privilege speed over the slower, more painful work of verification. Someone decided that the appearance of finished intelligence was more valuable than the texture of hesitant human thought. Someone, perhaps many someones, agreed that this was acceptable. And the social dimension of intelligence—the part that concerns itself with what is *shared* and *negotiated* among people—has been systematically eliminated from the process. What does accountability mean, then? Perhaps it means something like this: a willingness to slow down *after* the error is discovered, to trace backward through the decision-making process with the same care that one might read a difficult text. Not to assign blame, but to ask, *Where did speed substitute for thought?* It means, perhaps, insisting that certain decisions—decisions that affect many lives—retain within them some residue of slowness, some preservation of the hesitation that is actually the birthplace of responsibility. For intelligence without hesitation is not intelligence at all; it is merely fluency. And fluency, as any reader knows, can conceal a great emptiness. The error that surfaces three weeks later is not really an error in the machine's operation. It is an error in the *judgment* that permitted the machine to operate at all without human friction, without the social resistance that comes from having to explain oneself to another person, in time, with patience, as one does when one is genuinely accountable.