The Fantasy of Stylelessness

In virtually every guide to philosophical writing, there comes a point when the seasoned author implores the budding philosopher-writer to please not try to make their writing literary or elegant or sophisticated or distinctive or really anything that might draw attention to it as writing. “The purpose of (non-fiction) writing is to communicate,” claims Michael Huemer’s A Guide to Writing, “It is not to make art or to impress the reader with your sophistication.” The “First Commandment” of Adrian Piper’s Ten Commandments of Philosophical Writing is “Thou shalt not obscure thy ideas with turgid prose. In the Anglo-American analytic tradition of philosophical writing, we show respect for our ideas, our readers, and ourselves as committed intellectuals by attempting actual communication, not just verbal self-expression.” In Jim Pryor’s Guidelines on Writing a Philosophy Paper—my own entrée into the genre—the message is more plainly put (fittingly): “Don’t shoot for literary elegance. Use simple, straightforward prose. Keep your sentences and paragraphs short. Use familiar words. We’ll make fun of you if you use big words where simple words will do. Don’t write using prose you wouldn’t use in conversation: if you wouldn’t say it, don’t write it.”

When I think of the predicament in which academic philosophy finds itself today with the advent of large language models (LLMs), I think first of its indifference, bordering on contempt, toward writing as anything more than a means of information transmission. In the world of what I call “Points Philosophy,” the dominant model of academic philosophy today, the singular aim of philosophical writing (and indeed philosophy writ large) is to Make Points. Writing that is not obviously in service of that aim is distracting fluff. Even worse, it indicates something defective about its author: that they are engaged not in philosophy but in pretense. According to the doctrine of Points Philosophy, attention to writing as such can only be a mark of vanity: of “appearing to have a sophisticated writing style” or attempting to “wax profound.” A good writer of philosophy knows that you must “[u]se overly simple and concise language”; you must “[a]ssume your audience is extremely dull.”

The attitude is, of course, dubious on its face, since writing in philosophy and other humanistic disciplines could never merely serve to transmit results but is rather itself constitutive of the “result.” This is something that philosophers should appreciate better than most, given that the hallmark interventions of their trade are essentially based in attention to the subtleties of language—for example, intuitions about terms and concepts, interpretations of arguments and texts, carefully drawn distinctions. What’s more, throughout the history of philosophy, they have been embedded in a variety of literary forms, which are integral to the inquiry itself: Plato’s view of philosophy as fundamentally multivocal as captured in the dramatic structure of his dialogues; Augustine’s account of desire through exposés of his own in his Confessions; the aphorisms of Wittgenstein’s Philosophical Investigations, which comprise a therapy of grammatical “reminders.” And yet it remains nevertheless the case that philosophers these days seem to care little for how philosophical issues are themselves worked out in words.

This is not to say that contemporary analytic philosophy does not have a style. For his part, Manuel Vargas claims in his essay How to Write (Not Terrible) Philosophy Papers, “The house style of the discipline of philosophy is writing that is BORING.” But I would not simply call philosophy’s house style “boring” (though it often is that). If style is not just a matter of affect but also a marker of identity, then the dominant philosophical style today is stylelessness: a mode of writing that attempts at effacing the fact that a work has a creator at all. But while the fantasy that underlies stylelessness might have appeared fifty years ago as a mostly harmless expression of the philosophical aspiration to objectivity, it appears rather more ominous today. A technology which seems to speak from everywhere and nowhere on any issue, instantly, is now here. How can the Points Philosopher possibly resist?

Doing Philosophy

Indeed, in arguably only the first or second year of competent “writing” by LLMs, AI “assistance” in philosophy is already not merely permitted but warmly welcomed, even encouraged by many philosophers. Real data on usage is hard to come by, and most philosophers are concerned enough about their reputations to not publicly disclose their own AI practices, but find yourself among philosophers for long enough, and a picture of broad use will quickly become clear. Tell a philosopher that you’re currently doing any bit of work that is perceived to be lightly tedious and chances are that he’ll ask you why you aren’t using ChatGPT. Get to any bump in the road during a conversation, philosophical or not—Has anyone written on this? What did so-and-so say about such-and-such? How long does it take to drive to LaGuardia Airport?—and there is a high likelihood that someone will pause things to ask Claude.

It is uncomfortable against this backdrop to reveal that you do not, in fact, use AI to do philosophy. To say that you might not need it is to suggest hubris. To say that you positively do not want to use it is to invite charges of Luddism and moralism. To say that you just simply have not gotten around to incorporating it in your work is to look stupid and outmoded. A colleague of mine once compared my continuing to work without AI to an insistence on asking Jeeves, in the age of Google. Here is a tool brimming with competence and capability, just waiting to help answer my query, if only I’d get over myself and just let it.

The truth is, I do not use AI in my day-to-day work and life for mostly quotidian prosaic reasons. I have tried and found it not very helpful for the particular projects I am working on. I find that it sends me down unhelpful rabbit holes and wastes more of my time than it saves. I also don’t entirely trust its answers, so I anxiously check the veracity of its responses to even simple queries—a process which often becomes just as involved as that of simply going about tackling the original question directly. Most basically, though, I do not use AI, because I for the most part like the activities of my work enough that I do not feel the constant need to streamline it. Plus I find the experience of chatting with an AI to be tiresome and frankly, annoying. Its relentless peppiness. Its annoying cadence of “thought,” which delivers every response in a “Vox explainer” voice. Its obsequiousness makes me ashamed. Its longwindedness feels like filibustering.

To each of these reasons, the enthusiastic AI-user follows up. But have I tried the Pro version of ChatGPT? (Yes, I have.) Have I used Claude, because I really need to be using the latest Claude model. (No, I haven’t; maybe I will.) How long have I actually tried working with it? Because you do have to stick with it. It takes time to learn, just like any tool, but eventually it will be worth it. (I tried rather consistently for a couple of months, but it’s true that I can keep trying.)

Philosophers have a knack for missing the forest for the trees: In a time when OpenAI, Anthropic, and Google vie to hold the privileged role in active killing missions abroad and surveillance projects at home; when AI’s primary value proposition is the prospect of mass unemployment; when the founders of AI companies materially support an at-best proto-fascist presidential regime; when the entire enterprise of LLMs is built on the backs of staggering levels of copyright infringement; when the viability of AI depends on a Green New Deal in reverse, wherein the construction of highly materials- and energy-intensive data centers drives catastrophic levels of wealth inequality; when the public has lost trust in higher education in part, according to my institution, because universities no longer seem to be a place where students actually learn: What is the politics of eagerly embracing AI to do philosophy, of all things? And I do not just mean here, what is the ideological message conveyed by this posture towards AI? I mean also: What are the concrete practical effects of such vigorous proselytizing about the wonders of AI? The answer is surely in part: a contribution to and reinforcement of the hype that is structurally necessary for the investment cycles that keep the AI ecosystem afloat, despite the world-historical chasms between revenue and debt that characterize the economy and threaten its future viability. But I digress.

Back to the trees. For all the enthusiasm and angst about philosophy and AI, remarkably little is said about what philosophy is such that AI is good or bad at it. It helps then to ask the question explicitly: What exactly is this philosophy which is so easily able to be efficiently outsourced to AI? Most obviously, it is a vision of doing philosophy identified with its end-products (papers, arguments, objections, conclusions), rather than the actions that generate those products (writing, close reading, the development of a distinctive perspective). It is a view of philosophy as a body of information, and of “doing philosophy” as a sort of pattern-recognition project, a matter of making connections between different parts of that body of information. This is, at least, what is suggested by the kinds of anecdotes that are often offered as evidence of AI’s striking ability to do philosophy. In a recent article in The Chronicle of Higher Education, my colleague Daniel Greco explains that one indication that the age of AI philosophy is already upon us, is fact that AI can answer to Greco’s liking the “technically sophisticated and deeply unobvious question”—‘What challenges are there to simultaneously holding Ramsey-style analytic functionalism about mental categories and endorsing a theistic argument from design?’” I remain overall uninterested by how impressive the LLM’s response might be to a question like this; if philosophy amounts to playing language games with jargon, then we academic philosophers might as well pack things up and call it a day—surely, there must be something better for us to do.

More broadly, this vision does not resonate with why I do philosophy—and I would venture that it fails to capture why most people on reflection do philosophy. Philosophical problems have to be, in an important way, ours, in order to be able to set us off in search of their answers. But can philosophy really be mine, even as I offload more and more of my doing of it? A philosophy whose cravings are not mine is a philosophy whose satisfactions also cannot be mine. This view of philosophy can be accused of many things: it may be overly romantic, egotistical, frivolous, unduly skeptical or pessimistic about “getting things right.” But failing to have a coherent response to the massive changes wrought by AI is not one of them. According to this conception, the value of philosophy constitutively depends on human beings being authentically invested in its questions and taking ownership and responsibility over how we answer them. Philosophy is a distinctly human sickness and cure.

Who Do You Think You Are? Kant?

When philosophers do express concern about AI, they tend to focus on AI’s impact on the process of learning philosophy—the impact of AI in the classroom and on graduate training. Already, widespread use of AI has made it virtually impossible to assign students standard take-home essays. Nearly all of the assessments in my courses are now based on in-person assignments. Whatever take-home essays that I still assign are now coupled with oral exams. One obvious reason for these changes is to ensure that assessments continue to track reliable markers of student aptitude and mastery of material. But most philosophers (and teachers more generally) are concerned not just with rational schemes of evaluation but also with the effects of AI reliance on students’ cognitive capacities and development. Research has already shown AI usage to be associated with deteriorated memory recall, critical thinking, and basic skill acquisition. Offloading cognitive effort—surprise—turns out to leave a lasting effect on cognitive capacities.

And yet, from some philosophers’ heavy use of AI in their own work, it stands to reason that they are not similarly concerned with the effects of using AI on themselves, their capacities, and their philosophizing. Their divergent attitudes toward their students’ research practices and their own suggests that they believe there to be important differences between learning philosophy and doing philosophy. Using AI as a student of philosophy poses many risks to one’s ability to learn the activity, but once one is “fully fledged” as a philosopher, the risks are, apparently, greatly minimized.

There is no doubt some difference between the positions of students and teachers vis-à-vis the effects of AI use on users’ core knowledge, skills, and abilities. Failure to acquire and develop some skill at all due to excessive offloading of cognitive tasks to AI is surely more worrying than gradual skill degradation over time. I can also imagine that working philosophers are able to be more active in their interactions with AI, simply because they are more likely to have the expertise to challenge, probe, and redirect their AI interlocutors rather than passively taking them at their word. Still, I suspect that both the extent to which philosophers in fact use AI differently from their students and the extent to which they are insulated from the long-term cognitive effects of AI are easily overstated. Usage which is initially limited to certain tasks—tighten a sentence here, find a reference there—easily extends to more substantive requests for help. Every person who has used an LLM knows how eager it is to be at your service. Even setting aside its incorrigibly cheery offers to go beyond the call of duty (“If you share the full sentence, I can tune it so it sounds natural in context”; “If you want, I can put that idea in the tone of a philosophy research article or a more popular public-facing piece of writing”), there is also just the natural temptation to see whether it can do what you were about to. The persistent offers of assistance, just a few keystrokes away, embedded within the form of an ongoing dialogue draws you in. The difficulty in resisting this pull is universal—which is, of course, the point.

In any case, focusing on deskilling itself as the primary problem with using AI in doing philosophy is misleading, insofar as it detaches the skills and practices of philosophy from the activity of philosophy itself. But there is an intimate connection between that set of skills and practices and the actual doing of philosophy, such that, as teachers rightly worry, if you never acquire those skills, you have never learned how to do philosophy. The ability to transcend one’s own perspective, anticipate another’s, and strive through reasoned argument towards a more objective view of things is perhaps the very hallmark of philosophical thinking, such that if one never learns these skills and practices, then one never learned how to do philosophy.

But by the same token, if one is not actually engaging those skills and practices, because one has outsourced them, then one is not doing philosophy. Basic philosophical practices like locating yourself in logical space, reading others charitably, and handling objections, are not just practices of skill acquisition, merely preparatory to the subsequent doing of philosophy. They are constitutive of the very activity of philosophy. There is no asymmetry between learning and doing philosophy in this respect. What makes it the case that one is doing philosophy is not the mere possession of philosophical skill or theoretical ability to deploy it but its active use and practice. Offloading in this regard poses equal risk to the student and teacher of philosophy alike: the risk that you did not in fact end up doing philosophy at all.

At this point, I can practically hear the battery of questions from the philosopher reading this: Where is the line exactly, between the skills and tasks that are essential to the practice of philosophy and those that are not? And don’t we expect those skills and tasks to themselves be historically specific and so always changing? Would you decline to use a thesaurus? Are you going to stop talking shop with your colleagues? So you think it’s wrong to ask an undergraduate RA to do a lit review? And what if working with AI pushes you to be more active, think more intensely, and read more extensively than working without it? You think the right way to do philosophy is to just sit by yourself and scribble away alone, never seeking any kind of feedback from others? Who do you think you are? Kant?

Showing Up

The question of what it is to do philosophy is as old as the discipline itself, though the advent and eventual dominance of “analytic philosophy” in the twentieth century threw the divisions over how to answer the question into especially sharp relief. The philosopher most adept at speaking to these issues, to my mind, was Richard Rorty. Rorty was known for his own preferred conception of philosophy as an endless conversation, which he set against more “scientistic” approaches to philosophy that aspire to the loftier aims of truth and knowledge as modeled on the success of the (hard) sciences. Why, Rorty asked, must philosophy always strive toward a final answer, be it truth, reality, or God? Why must we conceive philosophical knowledge and success in terms of an ultimate terminus at all? For Rorty, philosophy should not posit “something nonhuman that human beings should try to live up to”; rather, he envisions a world where “human beings would wish only to live up to one another.” Living up to one another is noble and dignified insofar as we are ourselves noble and dignified. And yet it is hard not to feel somewhat let down by the idea that we should in the end only try to be as we have been, even at our best. Humanism feels small and parochial in a world so vast. Even the most stimulating of conversations eventually tires and tapers off. Shouldn’t philosophy of all things promise something more and greater than ourselves?

But when I think about the concrete experience of reading, writing, and doing philosophy, I am reminded of what it is concretely to strive to live up to one another. I think of the distinct feeling of being in the thrall of another person’s mind and the rush of gratitude that they had these thoughts and wrote these words, for me now to be able to take them in. These feelings of awe and appreciation are inextricably linked to the human recognition of what it must have taken for someone to write this: what they must have done, the time they must have spent, the strain and struggle, the activities foregone, the desires deferred, the times they nearly quit, the times they did quit and came back and started again. How astonishing we really are; and how exhilarating it is to be able to recognize each other in this way.

Even more fundamentally, the activities of reading and writing may themselves be understood as practices of living up to one another. When I write philosophy, I do not take myself to be merely informing you, the reader, of some Points, which are detached from me. Rather, my words are mine, and I assume responsibility for them. I take this to be the essence of our mutual contract, as reader and writer: I may expect your attention on what I write, only if I accept accountability for what I write and so, how I write. Rorty’s metaphor of conversation is especially apt in this respect. My part in our philosophical conversation cannot be efficiently outsourced, because doing so would undercut the basic ethical premise of being in conversation: that I show up, not in the sense of merely being physically present or nominally on the page, but in the distinctively normative sense of bringing myself fully to conversation and assuming responsibility for what I do in it. I offload what I think and what I say, then, at the risk of defeating my desire to really be in conversation with you: to hear what you say and respond, with my thoughts in my words, in turn.

To do philosophy is essentially to do it with others. It is also much else besides: trading reasons, constructing arguments, asking and answering impossible questions—even learning information, Making Points, and possibly “getting it right.” But its essential condition is that it is social. It emerges out of our being together: something which needs no help and admits of no improvement. To do it, we need only to do it.