I saw this article making its rounds on Facebook and LinkedIn as a provocative framing of academic publishing in the era of AI. Indeed, I appreciate Postdigital Science and Education for sharing banger after banger conceptualizing the relationship between AI, science, and education.
In this piece, Manabu argues that AI has reduced the labor value of the production of scholarly articles – articles long held to be the premier currency for academic success. To be fair, the value of traditional academic success metrics like publication counts, citations, or H-indices have long been in decay. Yet AI has now accelerated that inflation towards a precipice of obsolescence. This is because the machine now eats itself. It eats itself because AI stepped into a system of scholarly writing already digitized, modularized, and indexed for machine consumption — an AI-friendly tinderbox buffet. Manabu describes it thus,
The system that taught scholars to produce countable scholarly artefacts, rewarded them for producing those artefacts, and built its hierarchy on the scarcity of those artefacts cannot be saved by a filter once the artefacts become machine-producible. What is collapsing, in Bourdieusian terms, is therefore the entire field, not in the sense that scholars will cease to read and write, but in the sense that the specific configuration of capital, doxa, and legitimation that has organized academic life for forty years, no longer maps onto a coherent practice (Manabu, 2026, p. 13).
I love the way that Manabu connects the rise of AI in academic publishing to Bourdieu, though I quibble with the idea that scholars will not cease to read and write. Though “cease” carries extreme connotations, we already see evidence of declines in reading and writing at all levels of academia — from undergraduates upward. Even at the postdoctoral level, I notice many scholars framing AI as a beautiful shortcut to the taxing practice of reading or writing. As I enter my sixth year of graduate study, gnawing questions grow louder. Do scholars enjoy reading? Or do they enjoy extracting key information? For some, I suspect they are the same thing. In terms of writing, I suspect many academics appreciate the practice of having written – in which case, AI is the perfect antidote as one can “have written” thousands of words before starting their morning coffee. While that prickly procrastinative feeling Stephen Pressfield (2002) calls resistance used to be an everpresent foe for professionals who depend on writing for a living, AI has rendered resistance redundant. Why fight resistance when we can prompt and finish the task in a fraction of the time?
Through resistance, we can frame this quote from a phenomenological perspective. Until very recently, I took for granted that an academic article, an email, or any other text was written by a person – a person who may not write to me personally, but nonetheless deliberately chose their words, their cadence, their discourse. Sure, some genres have more mechanical or homogenized stylistic foundations. But it was still a human assessing their rhetorical decisions. Writing mechanically was a choice. That comfort is now gone. I find this unsettling because — as many LinkedIn conversations show (1) it is impossible to detect AI-generated text and (2) binarizing text into AI/human-generated is already the wrong assumption. To ask such a binary question is already to adopt the frame that both humans and AI generate text. Through the verb generate, we assume a frame of technological language even if we are talking about human creation for non-technological means. Furthermore, everybody’s interaction with AI is different. AI use exists on a multipolar spectrum and everyone’s prompting patterns constitute a unique AI fingerprint. Of course, agentive AI is now capable of self-prompting its own text. But that is not the world that I inhabit, and I will not talk about that here.
This dissolution of friction creates an alarming spiral of doubt. For graduate students and early career scholars, many of us ask: what activities would be most helpful to make me a strong job market candidate? We already know the job market is oversaturated. The numbers in my field of Applied Linguistics and TESOL range from 7-10% of PhDs who eventually land tenure-track jobs. Until recently I took for granted the value of writing academic articles, publishing them in Q1 journals, and being blessed enough for others to notice and cite those papers as well. As Manabu argues, this will soon no longer be the case. The best-case scenario is that this downfall of academic publication as a reliable indicator of scholarly ability will lead to more humanizing measures. Though any measure itself becomes vulnerable when it is leveraged as a metric. The worst-case scenario, my personal hell, is that nothing will unseat academic publications as the key marker of success, but the expectations of publication will rise to meet AI-inflected standards. Rather than making our lives easier, AI will make us busier than ever.
Now, I don’t want to be a Luddite and demonize AI-assisted writing. I use AI for speech-to-text dictation and writing feedback. Manabu even admitted to relying on AI to drafting his article – something I wondered as I read his liberal use of “not x but y” constructions. But as Manabu brings up in his conclusion, the greatest anxiety springs from the uncertainty of what flowers will spring forth after AI lays waste to traditional metrics of success. This is of little comfort to a current PhD student like myself, who constantly asks: “How do I stand out in a crowded field when that field is already collapsing in on itself?”
Yet I also appreciate how Manabu pushes back against the tiresome discussion of AI, plagiarism, and fraud. The transformative power of AI extends beyond that.
The claim is not that all AI-assisted scholarship is fraudulent, nor that human scholarly labour has ceased. The claim is that the legitimating mechanism, and the inferential chain from publication count to symbolic capital to social and material reward, depended on an assumption about the relation between artefact and labour that no longer securely holds (Manabu, 2026, p. 9).
The American Association for Applied Linguistics, AAAL, often points to published articles in high-impact journals as a key metric of success for postdoctoral scholars seeking employment and assistant professors seeking tenure. And as much as the field pays lip service to alternative forms of scholarly impact, nothing has come close to unseating scholarly publications as the driving factor. In some ways, the rise of AI is a welcome harbinger to the demise of publication counts, citations metrics, and their chokehold on measures of scholarly worth. Yet it also remains important for academics to publish academic texts, for other academics to read and engage with those texts in a documentable fashion. For this, citation is an efficient, if highly imperfect system.
I see AI usage, however, as a symptom of broader late capitalist cynicism — an ideology that prioritizes optimization and efficiency above slow engagement. This insidious ideology makes perfect sense for corporate interests, after Dodge v. Ford solidified the prioritization of shareholders above all else. And we also see individuals expressing this ideology via productivity hacks and self-improvement regimes – atomizing the individual as an entrepreneurial technology of the self. Sam Illingworth even shared an article that many university instructors are adapting to AI out of a resigned fear of falling behind rather than genuine excitement. Because, as the tired adage goes, “AI will not take your job, but someone adept at using AI will.”
Manabu convinces me of something I suspected: that AI levels a sledgehammer on an already brittle system of academic publishing and its corresponding metrics of merit. This parallels how reactionary, conservative, anti-intellectual regimes in many Western “democracies” have taken a sledgehammer to higher education. Part of me sympathizes with this backlash. Higher education in the U.S. changed forever in the mid-20th century – when the GI bill and the rapid rise of land-grant universities birthed an ideology that everyone ought to go to college. Within the liberal arts, I find it difficult to articulate the value of a college education for its current price tag. On the other hand, to argue that college is not for everybody is also to argue for exclusion and oppression. Yet the way that higher education has taken shape economically in the U.S. and elsewhere is clearly not sustainable. AI is just accelerating that demise.
Where will you go when the machine finishes ravaging the academic publishing field? For me, I do not know. I do not know if I will look back on my time in graduate school and regret my commitment to higher education. I do not know how I will make a living when making a living in higher education moves further out of reach. I may look back at my time in graduate school with regret — feeling like I wasted years of youth, only to slide deeper into jaded depression. Many scholars jackhammer the idea that the world is ending. And thanks to global inequalities, it is ending for some faster than others. If the world were already ending, maybe I would have been better off scrabbling together an ignorant, unlearned life. Maybe through ignorance, I would stress less. This would be true, if I treated my learning journey as a means to an end.
But instead, I come to appreciate the journey – a slog through the muck that can be managed rather than solved. I love reading not because it gives me the answers, but because it exposes me to new sides of the question. I write not because it will guarantee me a job, but because it gives me an expressive outlet. It gives me a sense of accomplishment to know regardless of how my academic journey ends, there will be a record that I was there – that I had ideas – and, more importantly, that I felt something along the way. AI leads us to obsess over output; it leads us to think of people as computer assistants; it leads us to think of language as a conduit of information.
But it is this conceptualization of language as a cognitive, informational medium that AI greedily consumes. The AI-human interface, however, breaks down when we think of language and creation beyond raw output. I notice it on YouTube when commenters complain about AI-generated thumbnails. If you notice that I use AI to generate thumbnails for LinkedIn and find that statement hypocritical, trust your instincts. Yet I still find the act of typing this sentence as its own form of slowing down – its own form of friction. And as Manabu (and physics professors) would argue, friction is a form of resistance. And resistance generates heat. And heat makes me feel alive.
References:
Manabu, T. (2026). The machine consumes itself: Artificial intelligence and the end of publish-or-perish. Postdigital Science and Education. https://doi.org/10.1007/s42438-026-00666-0
Pressfield, S. (2002). The war of art: Break through the blocks and win your inner creative battles. Black Irish Entertainment.
Dodge v. Ford Motor Co., 204 Mich. 459, 170 N.W. 668 (1919).
Leave a Reply