In 1964, when describing the threshold for obscenity in Jacobellis v. Ohio, the US Supreme Court Justice Potter Stewart said that while he couldn’t define precisely what constituted hard-core pornography, “I know it when I see it”. The problem, of course, is that the “I know it when I see it” test is highly subjective, because one man’s pornography is another’s erotic art. A similar issue plagues the subject of plagiarism. As Mario Biagioli (2012: 455) has argued, “Plagiarism is one of those notions that are stabilised by the emotions they express and elicit rather than by their conceptual clarity”. The very fact that it’s often difficult to reach agreement on whether plagiarism has or hasn’t occurred suggests that the topic is good to think with.
As someone with a longstanding interest in publishing and conceptions of authorship, it strikes me that the question of how ideas travel and who, if anyone, owns them, usefully highlights areas of tension between academic, trade and media publishing that require further unpacking in the post-human era of authorship in which we suddenly find ourselves. In what follows, I want to raise a larger set of issues about the somewhat uneasy relationship between academic writing and journalism, based on the different kinds of credit economies in which they operate, and what this means in the context of their increasingly symbiotic relationship and the rise of AI-powered language-processing models such as ChatGPT.
Parallels and interchanges between journalism and academia
Although journalism and academia operate in distinct social fields, there are numerous parallels between them that tend to disguise differences in their underlying models of authorship. First and foremost, they are highly desirable careers amongst people trained to think of the jobs as a calling – although in both professions employment is increasingly precarious, work conditions have deteriorated, and there are intense pressures to churn out “content”, and its academic equivalent, “outputs”, in a heavily metricised environment. Under such conditions, quality journalism and scholarship inevitably suffer, and sloppy practices and outright misconduct increase – as Gaming the Metrics (Biagioli and Lippman 2020) attests for the academic context.
In tandem with the decline of original and investigative reporting, repackaged content has arisen in its place. The practice has become so ubiquitous that it even has its own name: churnalism. Moreover, it’s hard to even blame journalists themselves for the practice, because the contemporary media publishing landscape survives primarily on the backs of poorly paid freelancers churning out content for clicks in the accelerated pace of a 24-hour news cycle.
However, not all journalists are precariously employed; nor do all newspapers churn out click-bait. As with academia, journalism has increasingly become a complex ecosystem in which a growing proportion of labour in any given organisation is outsourced to freelancers (“sessionals” and “visiting lecturers” in academic parlance), although reputable outlets typically employ a cadre of staff journalists, along with a few “stars”, to guarantee institutional credibility and accrue accolades. Like academia itself, the proportions of stars to freelancers varies dramatically from organisation to organisation, based largely on institutional capital and prestige. Indeed, much like Oxford and Cambridge (the same institutions that, not coincidentally, have produced many of their writers), stalwarts like the Economist and the Financial Times appear to have weathered the vagaries of the digital era better than initially promising upstarts like BuzzFeed and Vice Media.
There are also growing interchanges between the two sectors, with academics regularly contributing columns and articles to newspapers and magazines, or writing popular non-fiction, and journalists increasingly appointed into prominent roles at universities as chancellors and provosts. As Marilyn Strathern (2006) observes, key here has been the conceptual transformation of the university into a “knowledge producer”. With “knowledge” reconceptualised as “information”, this has led to growing demands for “expertise”. In this framework, academics are a group of “experts” waiting to be deployed – primarily by government and industry – to meet the needs of “society” (now inserted as an interlocutor, albeit sometimes via the rubric of the “taxpayer”). Thus, intellectual thought, now transformed into “expert knowledge”, must be publicly relevant, with academics incessantly exhorted to venture forth from the ivory tower under the banner of “public engagement”, “knowledge translation” and “impact”.
Conceptions of authorship in academia and journalism
But despite the parallels between academia and journalism, and their ever-tightening relationship, there remain significant differences that crystallise most tangibly in their contrasting conception of authorship. As Biagioli (2012: 454) has argued, “Authorship in literature or other works protected by copyright law is quite different from scientific or academic authorship”. What copyright protects is the form or expression of an idea: the words it contains, its diagrams, illustrations, etc. However, what is important to academics is the content, not its form: “the claims, the ideas, the results, the techniques”.
In essence, academics are not particularly interested in the property rights embedded in copyright itself, but credit rights, or attribution. This, of course, is precisely why we were so ready to relinquish our copyright to corporate publishers, with entirely predictable results (namely, they made a fortune selling our “property” back to our own institutions). It has also been a key impetus for the open access movement: we want our work to be circulated as widely as possible, without financial barriers to access.
It’s also why we haven’t traditionally claimed intellectual ownership over our course syllabi and teaching materials in the same way we do our publications. Although attitudes have changed in the wake of technologies that allow our teaching materials to be captured and reused (and even sold) by our institutions without our consent, they were historically conceived of as freely circulating gifts.
This feature of academic authorship makes it fundamentally different from authorship in other types of publishing – something recognised in the fact that academics retain copyright over their manuscripts rather than relinquishing it to their employer. Incidentally, this is why academic publishing agreements always ask whether you’re a government employee; in this instance, it’s the employer who owns copyright and assigns it to the publisher, not the employee. Academia presents a notable exception to the norm in this respect.
While attribution is obviously important in the context of trade publishing and journalism, what is ultimately at stake is royalties: the desired “credit” is as much financial as reputational. As Corynne McSherry (2003: 233) puts it, “Copyright law holds an infringer responsible whether or not there was any deliberate effort to deceive, because the copyright holder’s economic interest has been damaged”. For this reason, plagiarism is a fundamentally different type of offense in academia than in literary, trade and media publishing.
Defining plagiarism
Although plagiarism is considered a breach of ethics wherever it occurs, it is by far the worst of all academic sins. It’s why we watch for it so vigilantly in students’ work and take it so seriously when we spot it. The very fact that we have to spend so much time training students in what plagiarism is reveals that we are inculcating them into a very particular conception that is not remotely self-evident – as Who Owns This Text? (Haviland and Mullin 2009) illustrates at length. This is because in an academic context, it’s not royalties at risk, but something far more significant: reputations. To quote Biagioli (2012: 458), “Prosecuting a scientific plagiarist for copyright infringement would be like going after Al Capone for tax evasion”. According to Biagioli, this is why textual similarity alone is considered a more minor infringement than taking someone’s ideas and arguments and passing them off as one’s own.
In academic contexts, our ideas are fundamentally linked to textual sources: we warrant the former through the latter. These ideas are not primarily a result of the Muse, but of time and effort – labour, in so many words. We conduct empirical research, read widely in an area, come to grips with a range of sources, and then make a series of claims on the basis of what we have read and observed – much like pieces of a puzzle we have put together that could be arranged in multiple ways.
The success of the resultant “puzzle” is based on how well we’ve been able to convince our peers that the shape is pleasing and the pieces fit. Thus, our claims and our sources are inextricably entangled – arguably, over-entangled, given our intense fear of plagiarism and the excessive citation it tends to produce. (For example, so great is the fear of plagiarism that many academics, when suggesting that plagiarism is “good to think with”, would have cited Levi-Strauss, because the expression comes from him.) To use a different metaphor: we stake claims on the basis of sources, so we expect other academics to acknowledge our prior title deed.
I first had cause to think about this topic a few months ago, when Gillian Tett published an article in the Financial Times titled Brits keep washing machines in the kitchen. Americans don’t. Who’s right? The article bears a strong resemblance to one I published in June in Sapiens titled Do washing machines belong in kitchens? Many Brits say ‘yes’, which is a condensed extract from my book Silent but Deadly: The Underlying Cultural Patterns of Everyday Behaviour. Although she quotes me at length in the middle section, Tett does not mention my Sapiens article or book, or the extent of the inspiration they arguably provided in terms of her sources and claims.
But Tett is not an academic. Nor, in point of fact, is my article, or the book it is excerpted from, aimed at an academic audience. Indeed, I found the experience of writing a non-academic book disconcerting precisely because you don’t warrant claims in the same way – i.e., via constant citations. Popular non-fiction relies far more on the expertise and credibility of the author to warrant its content, and the author, in turn, doesn’t have the rhetorical security blanket that citations provide. In this context, name checks, bibliographies and hyperlinks more than suffice, and these are attended to with radically varying degrees of attentiveness from writer to writer – from impressively detailed to extremely perfunctory.
In fact, what academics see as plagiarism – i.e., what Biagioli (2012): terms “a displacement and recirculation of the plagiarised work in new different communities” – can be conceptualised as a form of “remediation” (Bolter and Grusin 2000) instead. Of course, that remediation generally serves to bolster the authority of the remediator rather than the originator, but it also gets ideas out into the public domain that would otherwise sit “uselessly” inside the academy. In effect, the process of remediation transforms intellectual knowledge into commodified information, and this transmutation entails a fundamentally different model of the author.
AI and the death of the author?
While a transformation of authorship is implicit in the shift from academic to public writing, there is still a designated author. Interestingly, this has remained the case even in the digital age, despite the theorising of media and literary scholars like Jay David Bolter (2001) and George Landow (1991) that hypertext would erase the figure of the author entirely by transforming an apparently stable, coherent text into a network of rhizomatic associations with no originary source. As Kathleen Fitzpatrick (2011) discusses, this is primarily because few humans have the time or the inclination to follow these associations, especially given the infinite potential of hyperlinks – of links within links within links connecting in an endless chain that leads the reader down an eternal rabbit hole.
Thus, to date, hyperlinks have not hastened the death of the author in the way that such theorists initially envisioned. It turns out that the very existence of such links, and the knowledge that we might follow them if we want to, means we mostly don’t feel the need to. In fact, if anything, the opposite has largely occurred: the sheer abundance of information available in the digital age means that many of us seek authoritative and trustworthy voices (i.e., “experts”) to mediate – or, at least, remediate it.
But that, of course, was before 2023. While humans don’t have the capacity to search a potentially infinite multiplicity of hyperlinks, AI language processing software does. And AI treats all those links as grist for its processing mill. Simply put, for ChatGPT there are no authors, only content. (So perhaps Jay David Bolter and George Landow were right after all – at least, taking the long view.) ChatGPT will happily produce a “hypothetical exploration” that is framed as an original analysis rather than a synthesis of existing information and that contains no sources, even if it is only drawing on one or two.
For example, if you ask ChatGPT how cultural differences might manifest in kitchen design and appliance placement, you’ll get a response that largely replicates the points of my piece and Tett’s, without citing either. In fact, as Ryan Anderson discusses in Anthro{dendum}, ChatGPT disclaims any responsibility for citation, noting, “I am an AI language model and I do not have the ability to cite sources… Without access to external sources, my responses are generated based on my pre-existing knowledge and training, which may not always be accurate or up-to-date”.
Who is plagiarising whom?
According to PC Guide, what ChatGPT is doing by collating information off the internet is not plagiarism. In a recent discussion of plagiarism and ChatGPT featured in the magazine, the author notes, “the language model does not directly plagiarise chunks of text that could be found elsewhere. Instead, it generates its own original content”. See the problem, academically speaking? The “original” content ChatGPT is generating is original in form only, not in actual content. However, as I’ve already discussed, this makes sense in terms of the main legal mechanism for determining plagiarism (at least in publishing): copyright, which is exclusively concerned with the economic interest of the author. But taken to its logical endpoint, as ChatGPT has done, means that anything and everything is original content, as long as it’s synthesised and paraphrased rather than quoted directly. ChatGPT is basically the ultimate remediator!
This is an aspect of ChatGPT that has received surprisingly little discussion amidst the flurry of cautionary pieces about the technology, although there are exceptions – such as Lincoln Michel’s recent article in Counter Craft: “The endgame for A.I. is clear: rip off everyone”. While writers are worried about being replaced by ChatGPT, lecturers are worried about their students submitting essays generated by the technology, and academics are worried about unscrupulous scholars cutting corners by using the technology, the primary underlying concern is that people will be tempted to ‘plagiarise’ from ChatGPT. That ChatGPT is effectively plagiarising the whole of the internet is largely absent from view. Moreover, invoking plagiarism in this way is treating ChatGPT as the ultimate author of the texts it produces, even though it explicitly states that it does not meet the criteria for such.
In sum, much deeper conversations are currently needed about what we mean by the concept of the “author” and whether accompanying concepts like plagiarism relate to intellectual influences, economic interests, or both. It seems to me that the traditional academic conception of the author, which has been increasingly discredited in an age of commodified knowledge and the concomitant rise of the public intellectual, offers us a way of understanding plagiarism that is particularly helpful in the age of AI. Because once we start prosecuting students for plagiarising ChatGPT, then we have essentially ceded the concept of authorship to the non-human synthesisers of the store of human intellectual thought, rather than the human originators of these ideas. And I’m pretty sure that for all those theorists predicting the demise of the author, this isn’t quite what they had in mind.
Acknowledgements: I am indebted to Ciara Kierans, Darlene McNaughton, Nadine Beckmann and especially Bruce Kapferer and Harry Rothman for their contributions to this article via critical readings. I also appreciate Gillian Tett’s willingness to engage with me about this topic, which informed my thinking about the differences between academic and journalistic conceptions of authorship. Finally, I would like to acknowledge Ian M. Cook and Sahana Udupa for their helpful feedback on the article following submission.
Academic references
Biagioli, Mario. 2012. “Recycling texts or stealing time? Plagiarism, authorship, and credit in science”. International Journal of Cultural Property 19, 453-476. (OA version available here)
Biagioli, Mario and Alexandra Lippman. 2020. Gaming the Metrics: Misconduct and Manipulation in Academic Research. Boston: MIT Press. (OA version available here)
Bolter, Jay David and Richard Grusin. 2000. Remediation: Understanding New Media. Boston: MIT Press. (OA version available here)
Bolter, Jay David. 2001. Writing Space: Computers, Hypertext, and the Remediation of Print. London: Lawrence Erlbaum Associates. (OA version available here)
Fitzpatrick, Kathleen. 2011. Planned Obsolescence: Publishing, Technology, and the Future of the Academy. New York: New York University Press. (OA version available here)
Haviland, Carol Petersen and Joan A. Mullin, eds. 2009. Who Owns This Text? Plagiarism, Authorship, and Disciplinary Cultures. Logan: Utah State University Press. (OA version available here)
Landow, George P. 1997. Hypertext 2.0. Baltimore: Johns Hopkins University Press. (OA version available here)
McSherry, Corynne. 2003. “Uncommon controversies: Legal mediations of gift and market models of authorship”. In Scientific Authorship: Credit and Intellectual Property in Science, edited by Mario Biagioli and Peter Galison, 225-250. London: Routledge.
Strathern, Marilyn. 2006. “A community of critics? Thoughts on new knowledge”. Journal of the Royal Anthropological Institute 12(1), 191-209.
