I’m an extremely fast and careless writer, which I suspect encourages me to be a fast and careless thinker. I often get carried away with a particular line of thought, more tied up in the satisfaction I’m finding in it than the disinterested evaluation of it. It’s just satisfying to express it once an idea is in my hands and I’ve begun the process of putting it into words. If we return to Sword’s (2023: loc 400) description of the pleasure of digital writing, it’s notable that she says “my physical surroundings seem to fade away” (emphasis added). The experience with typing often lends itself to this fading away of the context, at least if we are able to type with a certain level of fluency. The deeply engrained tendency to associate computing with virtuality plays a role in this, as a great deal of cultural baggage insists on trying to frame our digital experience as somehow separate from the material world (Carrigan and Fatsis 2021: ch 2). In reality, interacting with any computer is a deeply embodied process.
This is particularly pronounced for me at the moment, as I realize I’m hunched over my laptop on my kitchen table. My back is tired which is leading me to slump forward into a self-defeating posture which will in turn further tire out my back. I’m hunched over my kitchen table because I’ve spent too long today at my standing desk, which eventually makes my knees ache (even with the mat I bought to mitigate this) and involves too many repetitive motions which strain my right shoulder. But without defaulting to the standing desk, a whole series of other physical ailments reassert themselves for which the standing desk was (originally) the solution. These physical ailments, described by Till (2016) as the “industrial injuries of the academic” are common within higher education yet we are prone to seeing them as private matters, rather than expressions of shared working conditions.
The fact these tend to evaporate into discrete spheres of professional experience means that we often don’t recognize how many words we write each day across various activities. It’s often suggested that 1000 words a day is a sustainable target for academic authors. In my experience many academics who don’t have a formalized writing practice, particularly early career researchers, will express a sense of being slightly intimidated by this expectation. However, they are already typing thousands of words each day which are, in a real sense, original work even if it is not creative labor of the kind we dignify with the designation ‘writing’. To understand how you relate to writing it’s necessary to recover awareness of quite how ubiquitous writing, in this broadest sense, is within your working life. It is something we are doing constantly as academics, often across multiple devices and for a dizzying range of purposes, often without realizing we are doing it.
Once you start to recover a sense of writing’s ubiquity within your working life, it can be a curious experience to reflect with fresh eyes on this often tacit practice. The social theorist Nick Crossley (2013) captures something of this in his reflection on the phenomenology of typing. He highlights how our engagement in writing operates through embodied knowledge rather than a deliberate reflection on what we are doing. It rests upon a bedrock of habit which accounts for the specific experience of how we come to be immersed in writing or struggle to reach this state:
I can type and to that extent “I know” where the various letters are on the keyboard. I do not have to find the letters one by one, as when I first bought the thing. My fingers just move in the direction of the correct keys. Indeed, when I am in full flow, I seem actually to be thinking with my fingers in the respect that I do not know in advance of typing exactly what I will say. It is not just that I do not need to think about where the keys are, however. The break with reflective thought is more severe than this. I could not give a reflective, discursive account of the keyboard layout. I do not “know” where the keys are in a reflective sense and to make any half decent attempt at guessing I have to imagine I am typing and watch where my fingers head for when I come to the appropriate letter. The type of knowledge I have of the keyboard is a practical, embodied knowledge, quite remote and distant from discursive knowledge.
What he terms a remoteness from ‘discursive knowledge’ suggests an important feature of our writing. It’s not that we don’t stop to think when we write. In fact I stopped to think before I wrote these sentences because I understood the point I wanted to make but I wasn’t sure of the best way of making it. Once we identify this distance between the practical knowledge involved in typing and the discursive knowledge which is being drawn upon to give substance to the ensuing text, it highlights the different ways in which we can bridge that distance. Are we adequately caring for the stock of discursive knowledge upon which our writing ultimately depends? Do we ensure that our writing is ‘fueled’ by a rich stream of reading, listening, conversation and note-taking beyond the immediate practical requirements imposed by a specific text? Do we create the conditions in which those connections can gently unfold?
In his reflection on thinking, the philosopher Martin Heidegger drew an analogy between carpentry and less obviously embodied pursuits such as thinking and writing poetry. He sought to draw attention to how busy work, including the requirement to respond to organizational and economic obligations, closes down the space in which creative engagement with a task can take place. The relatedness to raw material which we are working with can easily be squeezed out by a preoccupation with outcomes:
A cabinetmaker’s apprentice, someone who is learning to build cabinets and the like, will serve as an example. His learning is not merely practice, to gain facility in the use of tools. Nor does he merely gather information about the customary forms of the things he is to build. If he is to become a true cabinetmaker, he makes himself answer and respond above all to the different kinds of wood and to the shapes slumbering within wood – to wood as it enters into man’s dwelling with all the hidden riches of its nature. In fact, this relatedness to wood is what maintains the whole craft. Without that relatedness, the craft will never be anything but empty busywork, any occupation with it will be determined exclusively by business concerns. Every handicraft, all human dealings are constantly in that danger. The writing of poetry is no more exempt from it than is thinking.
Many academic writers already struggle with ’empty busywork’ both within the writing process itself and in our wider professional lives. The enjoyment of writing depends on cultivating the relatedness to which Heidegger refers—to the ideas which, in the broadest sense, constitute the raw material of our work. Sennet (2008: 9) captures something important about this in his description of craft as “an enduring, basic human impulse, the desire to do a job well for its own sake” which “can be impaired by competitive pressures, by frustration, or by obsession.” The priority he gives to internal standards suggests the writing process is neater than in reality many of us experience it as being, even if we can nonetheless recognize what it means to be pleased with what we’ve done in a way that reflects no judgment other than our own.
This will always be a challenge to the extent that academic writing is a professionalized activity, as opposed to something we undertake in our time as a hobby. If our writing serves purposes extrinsic to ourselves then it will unavoidably involve negotiation between the satisfactions we find in it and our wider commitments and responsibilities. Finding a way to thrive at this intersection requires a degree of reflexivity about writing practice which graduate schools and existing forms of research training are ill-equipped to provide.
The increasing rate of digital change within higher education has meant that failing to cultivate this reflexivity is liable to produce practical problems for individual academics: as the technological landscape we are working within evolves, we need to reflect carefully on how we relate to it in order to realize the potential and avoid the associated pitfalls (Carrigan 2021, Carrigan and Fatsis 2021). This has been rendered urgent by the arrival of machine writing, with its tantalizing offer of the alleviation of the ‘busy work’ so inimical to a cultivated craft orientation to our writing process.
These remarkable new tools, such as ChatGPT and Claude, can enter into our practice in an almost infinitely flexible way. They have the capacity to be instruments of our craft, to support us in more thoughtful and creative ways of cultivating the relatedness to ideas which Heidegger invokes. But they can also be used to turbocharge our engagement with busy work, removing practical limitations on our capacity to meet obligations in a way which would have seemed like science fiction only a few years ago. They offer ways of negotiating the tension between internal and external demands which, if we’re not careful, could lead the former to be swamped by the latter. As we achieve more, we seek to achieve more. As we do more, we are asked to do more.
What I’m trying to raise are the downstream consequences of machine writing within organizations, where this might all be going in the coming years and what the habits we have picked up by then mean for our capacity for action. This is a real issue which is alive to me as someone who studied, supported and advocated for academics to embrace social media before becoming increasingly alarmed by how these platforms have evolved and the user culture they have given rise to (Carrigan and Fatsis 2021, Carrigan 2019: ch 6). How many academics would have thrown themselves into social media in the mid 2010s if they had understood what Twitter/X would look like by the mid 2020s?
What Doctorow (2023) has called the enshittification of social media, in which the initial promise tech firms make to consumers is ultimately swamped by their imperative to maximize profits, should lead us to be extremely cautious about what the platforms we commit to now might come to resemble by the time it’s too late to easily disentangle ourselves. It’s difficult but not impossible to stake out a middle ground between uncritical optimism and outright paranoia, in which we cautiously experiment informed by a reflective awareness of the still uncertain consequences (Carrigan and Sylvia 2021).
We need to attend to what might be lost in a careless embrace of machine writing by academics. I would argue that it is precisely the difficulty attached to the writing process, the confusion and frustration which every academic experiences at least some of the time, which we need to hang onto. It is this difficulty which machine writing offers an (apparent) solution to in the immediacy with which meaningful prose can be generated in response to almost any request.
The most recent generation of AI models have vastly surpassed the initial capacities of early versions like ChatGPT when it launched in November 2022. This creates a tendency for regular users and casual observers to have wildly differing understandings of what conversational agents can do. Even the firms themselves are unable to reliably specify what these models can do. Instead, they rely on the ‘collective intelligence’ of a user base to explain how these capacities play out in real-world contexts.
There are deeply strange features to conversational agents such as ChatGPT’s propensity to initially claim it cannot do something until it receives sufficient encouragement to try. “Go on ChatGPT, you can do it, I believe in you!” shouldn’t work as a mode for engaging with software but astonishingly it does in at least some cases. For this reason, I would encourage even the deeply skeptical to engage in a sustained way with these models, simply because it’s the only way to become meaningfully acquainted with their continually emerging capacities.
My belief that conversational agents could meaningfully contribute to academic tasks comes from my perspective as a daily user over the last two years. I was interested but skeptical until the launch of GPT-4. With Anthropic’s Claude 3, I was hooked. With ChatGPT-4o and Claude 3.5, I came to believe this was revolutionary technology, even if I remained uncomfortable with how my judgment would compound the hyped character of the technology.
The challenge ahead lies not in whether to use these technologies, they are becoming ubiquitous, but in how to preserve the craft of academic writing and thinking while engaging with tools that can so easily reduce our relationship with ideas to “empty busywork.” By maintaining a reflective awareness of how these tools shape our thinking and writing processes, we might find ways to harness their power while preserving what makes academic writing meaningful: that difficult but essential relatedness to ideas that transforms typing into thinking.
Mark Carrigan 