Deposited preprint
The withdrawal of the given
Philosophy as a way of life and what stands in for it
Cite this as
Montaruli, Alessio. “The withdrawal of the given: Philosophy as a way of life and what stands in for it.” Preprint, Version 12, 2026. DOI: 10.5281/zenodo.22735070.
This article is a companion to Reading Between the Times: Intus Legere, Vectorization, and the Structural Veil of AI Output (KiTalent Research, 2026; DOI 10.5281/zenodo.22735030), whose account of the text’s locus and of the structural veil its last section draws on. It was written with an AI assistant used as that section describes, and its historical claims were checked against primary sources. It prints the conditions under which its claims lose and asks to be engaged with, diverged from and tried on those conditions and on any a reader brings.
Abstract. Two pictures of philosophy pass for the thing itself: philosophy as a literary genre, and philosophy as a field of research. Both give recognition the authority to settle whether philosophy has occurred. This article says what philosophy is before it is either: a way of life that begins in the withdrawal of the given, the experience in which what one knows ceases to be exempt from the question. From that beginning it follows that the academic study of philosophy is a science of its object in the way military history is a science of war, and that it stands in for the practice whenever the deposit it preserves is held as the measure that need not answer; that the sophists’ promise, excellence acquired from those who teach it, is the structure of the modern department; that recognition which certifies is social while recognition which is philosophical does not certify; and that there is no peer review in philosophy, since a philosophy that emerges from the withdrawal of the given leaves no measure exempt from the question on which a peer could stand as its judge. Hegel on Kant is examined as the apparent counter-example and shown to state, in the Logic, the only deference a philosopher can pay: to enter the strength of the opponent. The last section places the language model among the routes by which philosophy is practised and shared. It is not someone and it is not nothing: it gives the tradition’s language back as relation, with the route to each source lost or falsely supplied. It can serve as a scribe within a limit and as the already-said made available for examination; what it cannot do is undergo the withdrawal of the given or hold the position from which a step is taken. The walk it requires is the return to the text’s position, and the discipline it requires is one sentence: the machine gives language, and the reader must restore time.
Two pictures of philosophy are so widely held that they pass for the thing itself. On the first, philosophy is a literary genre: a kind of text, with a recognised form, submitted to recognised venues and judged by recognised readers, so that whatever does not arrive as that kind of text is something else, noise, opinion, or a machine. On the second, philosophy is a field of research: a discipline with a literature, a set of ongoing conversations, a training, and a community whose acknowledgement certifies that philosophy has occurred. The two pictures agree on one thing, and it is the thing worth examining. Both give recognition the authority to settle whether philosophy is happening. A serious argument is one that serious people have received; an original theory is one the field has taken up; a philosopher is someone philosophers call one.
I came to state this because I was told, on a philosophy forum, that a person who uses an AI in their work should not bother trying to publish it, and that Kierkegaard was a silly example of philosophy outside the academy because the field has changed since he lived. I do not want to argue with the people who said these things. I want to say what picture of philosophy makes them sound reasonable, and why I think the picture is wrong.
The study of a thing is not the thing
The study of medieval history is a discipline. It has methods, journals, chairs, a literature the student must master, and standards by which work in it is judged. It is not the Middle Ages. The person living in what we call the medieval period did not live in the medieval period; there was no such period for him, and no chair from which to study it. The historian of the Napoleonic wars can know things about a battle that no man who fought in it knew, and the knowing produces historians, not generals. Nobody is confused about this in the case of history, because the object is safely in the past. It becomes confusing in the case of philosophy only because the discipline that studies philosophy is also called philosophy, and its practitioners, reasonably enough, call themselves philosophers.
The academic discipline is a study of philosophy: of its texts, its arguments, its history, its current problems, conducted with the rigour and the community that any study needs. That is a legitimate science, and as a science it is closer to botany or astrophysics than to the thing it studies. There is nothing wrong with that; botany is a fine thing to do. The difference is that the botanist’s science does not remove the plant, whereas the study of philosophy, when it makes acquisition of a body of results stand for the practice, removes the one thing philosophy begins from. To see why, one has to say what that thing is.
The starting experience
Socratic ignorance can be read as a rhetorical figure, a feint by which Plato’s Socrates disarms an interlocutor before the doctrine is brought in. That reading concerns what Plato does with the figure. The ignorance I am concerned with is a method, and it is the beginning philosophy requires: the refusal to accept anything as a neutral datum, a deposit, a result already in hand. Aristotle places wonder at the beginning of philosophising; Heidegger, at the end of his essay on technology, calls questioning the piety of thinking. I read these as routes towards the same starting experience: what was accepted ceases to settle what may be asked. That identity is the claim I make through them, not something their different words establish by themselves. Aristotle’s account proceeds from wonder towards knowledge of causes; my question is what becomes of that knowledge when it is once again held as a deposit. Whatever else a person has, erudition, intelligence, a doctorate, the approval of the learned, if he has not undergone the withdrawal of the given, he has not begun.
It is worth saying what that withdrawal is, because the phrase can be mistaken for something easier. It is not doubt of everything at once, and it is not accomplished by announcing that all assumptions are rejected; such an announcement leaves every operative assumption exactly where it was. It requires only, and exactly, that what one knows cease to be exempt from the question. An inherited distinction, an accepted account of nature, the very instrument by which one identifies an error: each of these can become something whose sense and warrant have to be reached rather than used. The given is not abolished. It stops being given.
Now look at how philosophy is taught. Students attend lectures in order to learn philosophy. The sentence is ordinary and it contains the whole misunderstanding. For it to make sense there has to be something called philosophy that is there to be learned: a body, a deposit, a stock held in reserve, guarded by authorised practitioners who admit new members, transmitted with exactness, such that a student who repeats it correctly is doing philosophy right and a student who does not, or who alters it while repeating it, is doing it wrong. Heidegger’s word for what technology makes of a river or a forest was standing reserve, a stock held ready for use. The word fits here too, and it comes from the same essay whose last sentence names the piety of thinking. Philosophy as standing reserve is the exact negation of philosophy as questioning, and it is the form in which the discipline has to hold its object in order to teach, examine and certify it. Preserving an articulation is necessary; without the deposit there would be nothing to return to. What turns preservation into the negation of the practice is holding the deposit as the measure that need not answer.
For the people who occupy the authorising positions in that system the experience is a heady one. To sit as the judge of what counts, to admit and exclude, to review one’s peers and grade one’s students, to be paid for it and deferred to, is very like being the voice of philosophy itself, and it would take an unusual person not to come to believe that these acts are what philosophy is. The students are encouraged in the same belief from the other side, because it lets them aim at something solid: a body of knowledge that effort can acquire and that a certificate can attest. Neither party is acting in bad faith. But what is being acquired and attested is standing within the study, and the study, precisely by making philosophy exact, correct and transmissible, trains the student to take as given what philosophy exists to withdraw. It is worth saying how, because exactness is also what philosophy demands, and making a distinction exact can expose what had escaped questioning. The word has two senses, and the study lives on the difference. In philosophy a distinction is exact when it answers to the matter: the phenomenon is the measure, and the deposit is corrected by it. In the study a statement is exact when it answers to the deposit: the examiner can recognise it as correct without walking its route, because the deposit is the measure, and that is the exactness an examination can certify, by comparison with the deposit and without walking the route. A student trained to the second exactness learns, by every mark he receives, that being right means matching what is given. That is not a misuse of exactness by careless teachers. It is what exactness has to mean wherever correctness must be recognised without being undergone. That is why it is not merely one science among the three but the furthest of them from its object: astrophysics does not remove the sky, and botany does not remove the plant, whereas the study of philosophy, done this way, removes the philosophical attitude and puts a curriculum in its place.
The sophists
The Greek word for what is being transmitted survives in an odd place. We say a thing is sophisticated when it is refined, technical, hard to master, and the word descends from the sophists, by way of a medieval Latin verb that meant to adulterate: for centuries a sophisticated thing was one that had been tampered with, and by the late nineteenth century the word had acquired a positive sense. Its ancestors are the sophists, whom our schoolbooks remember as the villains of Plato’s dialogues, sellers of rhetoric and relativism. That memory is mostly Plato’s polemic. The promise I am concerned with is exemplified by Protagoras: to teach, for a fee, the excellence by which a young man could conduct his affairs and act in the city. Its structure is that accomplishment can be acquired from those who possess and teach it. The word’s history recalls the problem; it does not prove that every sophist or every modern teacher occupies the same position.
The modern department does not sell rhetoric, but insofar as it offers mastery of the deposit as philosophy, its structure is that promise of the sophists and not Socratic ignorance. Philosophy, it says to the student, is sophisticated; you must be clever to learn it; you must acquire from your masters the craft of making it exact and precise so that it is correct; and you must then write it fluently, in English, in the recognised form, so that it is distributed and received. Every one of those is a genuine skill and every one of them can be taught. Socrates taught none of them, charged for none of them, and was put to death by a city that had a great deal of sophistication and did not know what to do with a man who claimed to know nothing.
What is happening, then
Philosophy, before it is a genre or a field, is a way of life: a practice a person undertakes, in which a question is lived with, from the withdrawal of the given, until something is seen, and in which the seeing changes how one goes on. Pierre Hadot spent a career showing that this is what the ancient schools were, and that their texts were exercises within the practice rather than finished objects for whose production the life was merely preparation. The variety of forms in which philosophy has reached us bears him out.
Socrates wrote no philosophy, and his not writing it is not exterior to his philosophy; it is his philosophy, the conviction, which Plato has him state in the Phaedrus, that the thing happens between two people in a place and that the written word cannot answer back. Aristotle survives largely in writings associated with teaching, compressed to the point where a sentence can produce a library, while the works he polished for readers survive only in fragments and reports. Origen composed by dictation to a team of more than seven amanuenses who relieved one another in shifts, with no fewer copyists behind them, the whole staff paid for by a patron. Thomas Aquinas dictated to several secretaries, reportedly to three and sometimes four at a time on different subjects; his own hand was called the littera inintelligibilis, the illegible script, and had to be transcribed by his secretaries before anyone else could read it, and what was transcribed, the quaestio in its technical Latin, demands years of specialised training before a page gives up its argument. The medieval universities did philosophy as disputation, recorded by others. Hegel wrote in a way that gave pleasure to no one, possibly not to Hegel. Kierkegaard published under invented names and was, in his own city, a figure of public quarrel and caricature long before he was a chapter in anyone’s syllabus. Wittgenstein published the Tractatus, later dictated to students, and left much of his later writing for others to prepare for print. Nietzsche wrote aphorisms and filled notebooks.
Nobody concludes from the dictation, the illegible hand, the unreadable prose, the pseudonyms or the notebooks that these people were not doing philosophy. The medium was whatever was to hand, and it has always changed: conversation, lecture, dictation, notes, letters, dialogue, treatise, aphorism, and now recorded lectures, video, audio, and text produced with a machine. These are not degrees of philosophy, from the peer-reviewed article at the top to the podcast at the bottom. They are ways of practising it and, more importantly, ways of sharing what it opens. A possibility one person has reached becomes available to another only through some such route, and the route does not decide the worth of what travels along it.
A way of life cannot be a discipline
If that is what philosophy is, then it cannot be a discipline or a field in the sense in which a syllabus defines what is to be acquired and an institution certifies the acquisition. A way of life cannot be contained in that office. What we call “philosophy” when we mean such a subject is the study of this way of life, and that study is a field in the ordinary sense, with everything a field has. The confusion is that the study has taken the name of what it studies and makes its own acts stand for the practice.
In Bananas (1971), Woody Allen’s Fielding Mellish regrets dropping out of college. A co-worker asks what he would have become if he had finished. “I don’t know. I was in the black studies program. By now, I could have been black.”
An archaeologist can reconstruct aspects of Etruscan trade or the chemistry of their bronze that no individual Etruscan knew. Even complete knowledge of that kind would not make him an Etruscan. What the archaeologist cannot do through that knowledge is live the Etruscan life he studies, and the reason is not a lack of information. Being an Etruscan was a way of living, and a way of living can be learned only by living it. The study of it, however complete, is a different activity, done by a different person, in a different life, and no accumulation of the one becomes the other. That is not a limitation of archaeology; it is what makes archaeology a science.
The same holds for philosophy, with one difference: the way of life is still available. The historian cannot enter the battle he studies, but the reader of a philosopher can begin to philosophise, in the seminar itself, on the page in front of him, at the moment a settled distinction stops holding. It can be lived now, by anyone, and it does not care about any of the things the study is made of. Grades, examinations, admissions, tuition, salaries, scholarships, conferences, journals, chairs: these can open or obstruct the routes, and the time, through which someone comes to the practice, and that is not nothing. But none of them touches the practice itself, because none of them is a way of living with a question. A philosopher may take part in all of them. He may need to, in order to eat, and there is nothing dishonourable in that. He may even say, truthfully, that he does it in order to bring others to the way of living, and he may succeed in showing it, by example, in a seminar room or on a page, to someone who was not looking for it. But the person who attends his lessons will not acquire the way of living merely by mastering the lessons. He will learn it only by living it, and if instead he studies it as a subject, masters its literature, passes its examinations and is admitted to its guild, he will have done something real and will have missed the entire field. The professor who knows this teaches with it in mind. The one who has forgotten it teaches Etruscology and calls it being Etruscan.
Recognition
It follows that recognition, in the sense of admission and standing, is a social phenomenon, not a philosophical one. It can occur, and when it occurs it is welcome, because it widens the route by which a possibility travels. But it does not make a philosophy real, because a philosophy is not the kind of thing that becomes real when others accept it. It is practised or it is not. Kierkegaard’s was not less philosophy in the decades when Copenhagen took him for a crank, and it did not become philosophy when the professors arrived.
There is another recognition, and it is philosophical. A reader who comes back with a consequence the author did not foresee, or a difficulty the work cannot answer, has recognised the work in the only way that counts, by doing philosophy on it, and the work owes him an answer. But that recognition certifies nothing, and nobody needs it as a certificate, because the reader who can give it is already inside the practice, and the practice does not wait for him. So the recognition that certifies is social, and the recognition that is philosophical does not certify. Education, intelligence, erudition and the acknowledgement of one’s peers are signs of something, and what they are signs of is standing within the study. They cannot settle whether anyone has lived with a question until something was seen. That can happen in a professor and in a person who has never entered a university, and the study has no instrument for detecting it in either, because it is not the kind of event the study is designed to record.
This is also why the demand to state one’s idea in two sentences, before any context, is not a test of philosophy but a test of genre. A sentence of philosophy, detached from any route into the position from which it was reached, cannot by itself supply that position. “For Heidegger being is understood through time” names a direction without giving a person who has not walked it the position it names, and the same is true of any sentence that has been earned. The two sentences can be given; they cannot be assessed by someone who insists on not walking.
There is no peer review in philosophy
The strongest form of the recognition picture is peer review, and it is worth saying plainly that there is no such thing in philosophy, and never has been.
By peer review I mean the office that claims authority to judge a philosophy by its fit with an authorised common measure and to decide its admission on that basis. Hegel does not review Kant in that sense. He writes about him, at length and often harshly, but that is commentary, in the sense in which Avicenna comments on Aristotle: a thinker taking the measure of another by thinking his thought through and out the other side, owing him everything and diverging from him at the root. Commentary claims no authority from an office. The commentator has to earn every finding through the position; he does not decide whether Aristotle may be admitted, and the idea that he might would have struck him as a category mistake. Aristotle does not review Plato; he spends a lifetime engaged with him and leaves the Academy. When a philosophy emerges from the withdrawal of the given, it leaves no common ground exempt from the question on which a peer could stand as its judge. What has been withdrawn is the ground’s authority to settle the question without answering for itself. Shared words and distinctions remain; their being shared does not give them jurisdiction. What the new emergence does leave is a ground for those who come after and take it as given: its scholastics, in the literal sense, the school that studies it. They can review one another, because they share the deposit as their measure. The philosophy itself refuses that exemption, and the nearer a philosopher stands to his own scholastics in granting it, the less he is still a philosopher and the more he is the first professor of himself.
None of this makes philosophy unaccountable. A philosopher engages his peers with respect and deference, learns from them, and diverges from them, and the divergence is the engagement. Two things are owed here and they are owed to two different addressees. Deference is owed to the person, the one who lived from a question and paid for what he saw. Engagement is owed to the position, and it has no limit: a position is there to be thought through, and thinking it through may leave nothing of it standing. Hegel, who speaks of der bornierte Verstand in his criticism of Kant and calls Kant’s exposition barbarous, looks like the counter-example, and he is not, because for a thinker of his kind engaging the position without reserve is the way deference to the person is paid. He states it himself in the Logic: a true refutation must enter into the strength of the opponent and place itself within the circle of that strength; to attack him where he is not, and win there, advances nothing. The only refutation of Spinoza, he says, is first to recognise his standpoint as essential and necessary, and then to raise it, out of itself, to a higher one. That is what he does with Kant, whose unity of apperception he names among the deepest and most correct insights of the Critique before he takes it further than Kant would go. Whether a harsh word hits the position or the man is a question of reach like any other, and since how a thinker writes a distinction belongs to what he thinks, the line is not where politeness draws it. Where criticism turns into contempt for the person it falls short of the deference owed, and no achievement licenses the turn. The deference itself is in the whole of the work, and it is paid to the man: it consists in living from another’s thought and answering for where it leads. It is also a form of self-respect, because it tells the truth about where one stands.
A philosopher today can also do something harder than review: he can print, with each claim, the conditions under which it loses, and invite anyone to run them. A reader who runs those trials and reports where a claim fell has done something real, and the philosopher owes him the result. Nor is the printed list the boundary of what such a reader may do. He may find that a trial presupposes what it should test, or that its declared defeat would not touch the claim it purports to expose, and that finding is owed the same answer, because the conditions are themselves claims and are as losable as anything they test. The reader has brought a position of his own; without one he could not walk the route at all. What he has not earned merely by bringing it is an exempt measure over the work. His finding must show how it reaches the position, and his own measure must answer where the finding depends on it. He has walked the route and come back with a finding, which is what the philosophy asked for. If he writes that finding in a referee report, it keeps its force. The force comes from the walk and the finding, not from the reviewing office: the office neither supplies it nor gains philosophical jurisdiction by receiving it.
There is a second thing a reader can do, and it is where criticism can go wrong even when its local result is correct. He can show that a premise is false. Suppose he disposes of an ancient account of nature built on four elements by appeal to modern physics. He may be entirely right about the composition of matter, and the correction may remove a premise on which further claims depended; none of that has to be taken back. But what has he engaged with? He has extracted a proposition, translated it into the terms of a later science, and answered it there. The question the account was carrying, what nature, or change, or a beginning meant in the philosophy he is criticising, has not been met; it has been silently replaced by a question the later science knows how to answer. A verdict can be correct and miss the philosophy entirely. This gives the philosopher no shelter, since he must in turn say what the correction changes and how far it reaches, and he may find that it reaches further than he would like. What it asks of both is proportion between a finding and what is claimed for it. A local refutation is owed its force. A claim to have disposed of a philosophy is owed an account of how that force reaches the philosophy. The reviewer who supplies the first and claims the second has earned the local result and not the disposal of the philosophy. In passing from one to the other without the missing account, he has changed the subject and called the change a result.
What no peer can do is sit in judgement over a philosophy from an outside position whose authority is exempt from the question. He can bring his position into relation with it, and that relation may undo either position; he cannot make his standing outside it a title to judge. So when a reviewer invokes the office in place of the missing walk, that invocation tells you about the authority he claims, not about what the philosophy has reached. A peer is one who does not exempt his own position from this demand. Insofar as a review claims that exemption, the review itself shows that the reviewer is not acting as one. Disagreeing with this article does not claim the exemption. Refusing to answer for the measure by which it is judged does.
That is the reception any philosophy should expect and the only one it can honestly want: to be acknowledged, engaged with, diverged from, and tried, including in the conditions it states; never assessed from a measure held exempt from the question. Everything else that happens to it, the grades, the citations, the locked threads, the chairs, is the study doing what the study does, and it is happening somewhere else.
Where the machine belongs
In this picture an AI is one more route, and the question about it is the question asked of every route: what can be reached through it, and can what is reached travel along it to someone else? To answer it one has to say plainly what comes out of the machine. It is not someone and it is not nothing. It is the language of the tradition, gathered from millions of pages, broken into pieces and held as relations among the pieces, and given back as fluent text in the shape of an answer. A generated sentence need not have been written before; the relations can be recombined into something new. But every piece once came from somewhere: a person, a time, a dispute, a book written under a particular pressure. What the machine keeps is how the pieces relate. What it does not keep, as such, is where each of them came from and what it cost to say it: a relation can be preserved and transformed while the route back to its sources is lost, or supplied falsely, or recoverable only by work the output does not do. So a generated sentence can be correct and still belong to no one: no one has undergone the judgement its form appears to report, and recovering its sources afterwards would repair a loss of evidence without supplying that missing passage. That is not a fault to be engineered away; it is what the thing is, and it is also why it is useful.
It can be used as a scribe, within a limit. Dictation is old: Origen and Aquinas spoke and others wrote, and the work, the finding of the sentence, happened in the one who spoke. A generated draft is different, because it can supply the formulation itself. Where accepting it replaces the work of finding what one means, that work goes unperformed, and a writer who lets it be replaced every time has given up the one exercise by which a sentence becomes a discovery, since one discovers what one means in the act of writing it and not before, and an exercise never performed lapses like any other. So the rule is not “delegate the draft” but “know which drafting is the work”. Plans, records, apparatus, a first form to argue against: these the machine may draft, and I correct them and own them. But a corrected draft is not yet mine. Improving a sentence is editing; what makes a sentence mine is that I can answer for it. That is the difference between a scribe and a ghost-writer.
It can also be used as something I will not call an interlocutor, because no one is there in the sense a conversation requires: it answers without risking anything, revises without having been wrong, and can be redirected to the point, but no one is there who left it. What it does is more modest and, for a philosopher, more valuable. It makes the tradition available. When I think aloud against it, a claim of mine comes back with possible predecessors, neighbouring disputes and refusals, in seconds, as a map of the neighbourhood the claim lives in. That is a real gift; it is the first gathering from which reading begins, and no library, colleague or memory does it at the speed of a sentence. It is also the already-said, inherited language rendered and recombined, handed back to me as if it were a personal answer, and that is exactly what Socrates needed his interlocutors to supply, so that it could be examined. The sentence I have to write in order to refuse it is very often the sentence the work needed.
What happens in that exchange should be stated exactly, because it is where the accusation lives. The position I am speaking from is asserted and reasserted: I refuse a rendering, insist on a distinction it has lost, discover that my own formulation lets something through that I cannot accept. Through that work the position is refined, clarified and more exactly distinguished. It is not born there. The speaking already comes from somewhere, and the machine’s words cannot supply the standing from which I question them. This does not require that every useful sentence originate at my keyboard, or that I arrive with a finished doctrine. A formulation, a distinction, a consequence I had not articulated can emerge in the exchange, and sometimes does. But its becoming something I see requires a taking-up that the generated answer cannot perform, and until that has happened it is a formulation I have been given, not a thing I have seen. The position is clarified in the work, not manufactured by the answer.
But available is not present. What the machine returns has been rendered, not encountered. A hard, old, disputed text comes back as smooth present-day prose, with its distance removed, and distance belongs to the work of understanding a text; removing its signs does not accomplish the crossing. So the machine gives the map, and there has to be a walk, and one must be exact about where the walk goes. Not to the book as an object on a shelf. To what the book was written from: the question its author was asking, the time and the dispute he was asking it in, the things he could not say and the things he did not need to. I call that the text’s locus or position, the place from which its claims could be claimed. The book is one road to that position, and its words can resist you on the way without adapting themselves to your demand. That resistance can expose a wrong route; difficulty alone does not prove that the route is right. The machine’s rendering offers another road towards the position. It can contradict me, but its contradiction is not the independent resistance of the source: the next prompt can obtain a different rendering while the source remains unread. Neither road is the position.
And here is the consequence, which cuts both ways. Whether a rendering has rendered a position at all, or the book’s position rather than a plausible neighbour of it, can be judged only by someone who already has a position to judge it from. No reading happens from nowhere; the reader’s own standpoint is not a bias to be eliminated but the condition of reading anything. So a person who can orient himself, who knows from his own practice what a position is and when one is being described, can trust the machine’s rendering without reading the book, as an act of judgement for which he answers. What that judgement buys should be stated exactly, because it is less than it feels like. It does not create the evidence it lacks. A formed reader can recognise a coherent position and still be accepting a plausible neighbour of the one the book holds, and the warrant he needs rises with what he asks the rendering to bear: for provisional orientation, he can proceed on that judgement; where an attribution, a quotation or a claim of priority carries the argument, there has to be a return to something capable of resisting the rendering. The book, manuscript, letter or other primary record relevant to the claim must be allowed to resist. And a person who cannot orient himself is not saved by reading the book, because what he carries out of it is a summary in his own head, and the problem has simply moved indoors. What makes the return possible is not the paper but the place one returns from.
A general deciding from a map of country he has never seen is in this respect in the same position. He can use the map because he has walked other country: he knows what a river does to a column, what a map can show and what it cannot, why a blank on the map cannot be trusted to mean flat ground. A man who had never stood on ground could not read the map at all; he would see lines. And the map does not decide. The general decides, from the map, and answers for the battle; the map answers for nothing, and he does well to remember who drew it and for whom. That is the machine’s rendering of a position, and that is what it is to trust one without having been there. Ask the machine to object, and it will object, fluently; the objection is a lead to be tried, not evidence that a trial has been run, and a person who only ever asks the machine to criticise itself has not left the map. And the general who relies on maps alone, and has never seen the territory, will soon learn, at a price, that the judgement needed to read a map properly comes only from the field. The price is paid on the ground; it is never paid by the map.
What the machine does not supply is the thing itself. I am speaking of the present language-generating systems: no manner-world is disclosed there in which something matters for the system, no own having-been from which it answers. Calling this worldless does not place it outside the world in which we live or outside what happens. The generation is a real technical occurrence, and its reception can alter the person using it. What it does not establish is the lived position implied by the voice. The withdrawal of the given requires a world in which something can cease to hold as given for someone; producing its verbal account does not undergo it. When the machine produces a sentence in the shape of a judgement, the production has occurred, but that is not evidence that the judgement was undergone. The danger is not chiefly that the answer is wrong. Error can be checked, though it may pass unnoticed. The deeper hazard is a successful appearance, strongest when the output is good, and it is completed on our side, because we are trained by speech itself to hear an answer as the trace of someone who can answer for it. The person with nothing behind him is not deceived by the machine. Its words become significant only within his life, and it is there, out of his own need, that he lends the voice a speaker who could answer for it, and receives an eloquence with nothing behind it as if it were the end of the work, when he is only after the prompt.
The remedy, then, is not to refuse the tool but to refuse its magic. Its usefulness is real and its independence is a fiction. Concretely: keep what is settled written down where a conversation cannot silently amend it, without exempting it from being reopened; count nothing the machine says as the agreement of another thinker, and treat its objections as leads; send every strong claim out with the statement of what would defeat it; and, above all, return. Return to my own question before I return to the prompt, because a question ripens and a prompt only elicits. Return the answer to the position it came from, in proportion to what rides on it, and let that position resist the summary. And return the machine itself to its place, since the voice that speaks as if from nowhere is owned, powered, cooled and contracted somewhere, with shareholders and terms of service, and a philosophy that refuses the deposit’s gatekeepers should not kneel to a new one because it answers politely. A procedure can record these returns; it cannot undertake them. They require a discipline in the sense of a sustained practice, learned by being lived. Its whole content is one sentence: the machine gives language, and the reader must restore time, that is, must put back where each thing came from, what it cost, and who answers for it now.
There is, finally, a test of who is thinking in that arrangement, and it concerns where a step is taken, not whether a file is unique. Everyone has the machine. The same models, trained on the same libraries, are open to every student, every professor and every crank, and have been for years. Different people get different outputs, of course, since a prompt is already a choice, and the uniqueness of a file proves nothing about who thought. The point is where a step comes from, and there is an exercise a reader can run. Ask any of these systems, cold, for a category of being the tradition does not have. When I asked, it answered from Aristotle and Kant, which is to say it answered no, politely, with references; I do not promise that every first answer will, and a reader who gets a candidate instead should look at whether it is a category or a name for one. What the exercise shows is not who can take a step. No output can show that, since the whole point is that a formulation is not a step. What it shows is what the machine gives to everyone: the tradition, as relation, in the form of an answer made from the tradition’s language, faithful to it or not, which is a further thing to be tried. What it gives to no one is a step the tradition has not taken, and the reason is not that it cannot produce new sentences, which it can, but that a step is taken from a position and a position is what it does not hold. If such a step has been taken, it was taken by the person at the keyboard. In my case it was taken against the machine’s first answer; it could as well be taken by taking up something the first answer supplied, since what matters is the taking-up and not the order of events. In my case, the prompt that made a second answer possible was already a position, whose force lay not in its being a prompt, since a prompt can be copied or generated without anyone holding what it says, but in the insistence I could sustain and the distinction I had to answer for. The machine can supply new language, propose a connection, or make something already said newly available. It does not thereby hold the position from which the proposal becomes a question, a discovery or a refusal. So the test to return to in each exchange is this: what have I actually seen and taken up, and what remains a formulation I have been given? The machine’s contribution belongs in the account of how the work was made; it is not evidence that the passage occurred. That is what I meant when I wrote, in the Personal Note in Reading Between the Times, that the instrument helped me reach the giants faster and did not do the thinking, and that the difference between the two is part of what the book is about.
What philosophy opens
None of this says that philosophy has no results, or that anything a person lives with counts. It says where the results belong. Ontology, the science of being, is not philosophy in the sense argued here. It is what philosophy, as a way of life, can open an access to. The results are real in the sense that results are: they can be stated, tried, and defeated, and a discipline can study them. The way of life that reached them is not real in that sense and does not need to be. It is possibility held open. The life through which the results were reached occurred once, and no later reader can make that occurrence his own; what he can take up is the possibility, which none of the lives that held it open exhausts, and to take it up is to live it again, not to restore the first. It is lived, by someone, from the withdrawal of the given, with whatever route was to hand, and the forum, the journal, the citation index and the machine are all inside that life or they are nothing.
That is why the medium is not the criterion, why the field is not the practice, and why institutional recognition, welcome as it is, arrives from outside the thing it recognises. The study of philosophy will go on producing excellent scholars of it, as the study of war produces excellent historians. Whether anyone is doing philosophy is a different question, asked in a different place, and it has never been answered by a vote.
Which is why this article cannot be peer reviewed. It can be read, engaged with, diverged from, and tried, on the conditions it states and on any a reader brings, and a finding is owed an answer wherever it is printed. Anyone who reviews it instead will have said something, and it will not be about the article.