GenAI-Generated Authorship: “Where Does the Danger Lie?”¹

ABSTRACT

The discussion in this text is based on the assumption that epistemic authorship is, without a doubt, a scientific demand, made possible in and by scientific discursive practices, which goes beyond a richer vocabulary, a more articulated syntax, and more complex texts - generated by GenAI-Generated -, as its creators would have us believe. None of this corresponds to the “real” of language of language and scientific discourse in the production of meaning, as we aim to discuss below, from the perspective French school of discourse analysis. Thus, we intend to problematize the concepts of scientific writing and its correlate, the scientific text, the epistemic author-function, and also its correlate, the reader-effect, which have been guiding practices of this writing through the generation of texts by GenAI-Generated, based on prompts - texts that describe the task that the GenAI-Generated must perform.

KEYWORDS:
Generative AI; Epistemic Authorship; Discourse and Power

RESUMO

A discussão empreendida neste texto parte do pressuposto de que a posição-autor epistêmico, é, sem dúvida, uma demanda científica, possibilitada nas e pelas práticas discursivas científicas, que vai além de um vocabulário mais rico, de uma sintaxe mais articulada, de textos mais complexos - gerados por Inteligência Artificial Generativa (IAG) -, como desejam fazer acreditar seus criadores. Nada disso corresponde ao real da língua e do discurso científico na produção do sentido como objetivamos discutir adiante, sob a perspectiva da análise do discurso de orientação francesa. Assim, pretendemos problematizar os conceitos de escrita científica e de seu correlato, o texto científico, da função-autor epistêmico e, também, de seu correlato, o efeito-leitor, que vêm balizando práticas dessa escrita por meio de geração de textos por IAG, a partir de prompts - textos que descrevem a tarefa que a IAG deve executar.

PALAVRAS-CHAVE:
IAG; Autoria epistêmica; Discurso e poder

...under what conditions and through what forms can an entity like the subject appear in the order of discourse…[?]¹

Michel Foucault

What is an author? This is a complex question, as Foucault clearly demonstrated in at least three of his works, namely: What is an author?, a lecture given on February 22, 1969, at a meeting of the French Society of Philosophy; The Archaeology of Knowledge, published in March 1969; and The Order of Discourse, his inaugural lecture at the Collège de France, delivered on December 2, 1970. In these works, the philosopher reflects on the categories of author, authorship, and author function, in a movement that seeks to criticize the principle of universality of authorship as a fundamental and creative instance of discourse, which has dominated the traditional history of ideas (Foucault, 1981).² What he highlights in these studies is that this principle does not prevail everywhere or constantly, in all socio-historical contexts and discursive orders. In other words, each discursive order, whether literary or scientific, to remain within the domains explored by Foucault, establishes who has the right to speak, defining the status of authorship.

Given this complexity, in this work, our attention turns to questioning who this subject is and what their position would be in the order of writing practices presided over by generative artificial intelligence (AIG).³

This is a necessary question - widely debated in various fields of knowledge such as law, engineering, philosophy, and linguistics⁴ - and we believe it is more urgent than ever. This is because, with the advent of this new digital technology, especially that applied to the generation of written texts- Zotero, ChatGPT, Google Gemini, and Jasper, to name just a few of its models- and in view of its widespread use,⁵ we are witnessing a change in writing gestures and discursive practices. This change is evident in the fact that AIG, in its set of functionalities, has the ability to autonomously generate new content- text, images, video- in real time, in the terminology of the platforms. When it comes to writing texts, this digital tool, forged in complex algorithmic systems, operating through the statistical recombination of large volumes of textual data and predicting word sequences based on recurring patterns, dispenses with the subject’s writing work. The subject is put on hold. To create the text, the program user simply requests the AIG platform to do so, according to their production project, through so-called prompts - “textual instructions provided to a language model (LLM) to request responses or generate specific content” (Scalco, n.n.).⁶ In short, the text produced is entrusted to a programmed automaton, to algorithms, as these are trained to imitate human language, simulating the textual forms of written discourse historically constructed in the social body of literate cultures.

We are facing a technological advance, which, however, comes with challenges, as we admit that, in the field of scientific writing (and not only therein), meanings and practices are (de)stabilizing, in one way or another, the subject’s relationship with writing, with the ways in which they signify themselves in it and signify it, with the ways of producing text, with the modes of material production of knowledge. All of this, as Orlandi (1986) warned us when reflecting on the presence of technologies in the functioning of writing and reading practices, will consequently imply the “reorganization of intellectual life, redistribution of places of interpretation, promoting a shift in the functioning of authorship and the very conception of text” (Orlandi, 1986, pp. 62-63).⁷ And, along with this, issues that touch on the sphere of ethics, copyright, and intellectual property, for example, which have been used to characterize the world of writing, are also in flux (Chartier, 2022). Turning back to Foucault, given that we are dealing with a shift in the order of the written word, it is in this sense that his treatment of the figure of the author in the works cited interests us, bearing in mind that the relationship between the subject, discourse, and power constitutes one of his possible positions.

To advance this discussion, it is important to note that this paper does not aim to provide answers to the above questions, as the most challenging aspect has been understanding the transitivity of these categories in recent years (see, for example, Lana, 2023). Our intention here is to broaden this debate, seeking to build an understanding of them. We do so by mobilizing concepts from French-oriented Discourse Analysis, a theoretical perspective that leads us to work on the point of articulation between language and ideology in discourse, opening up the possibility of scrutinizing the way in which meanings and subjects are produced, observing the facts of language and its practices in the midst of capitalist social formation, as Pêcheux (1995) reminds us.

To this end, as a preliminary approach, we intend to suspend the widespread use of AIG features linked to academic and scientific practices, which include, for example, searching social networks for guidance on research writing and the use of AIG models for searching and generating scientific texts, which consist of “supporting,” through “autonomy,” university students in data management and decision-making (Lin; Chang, 2020).⁸

We understand, from the theoretical perspective adopted here, that this “training” for writing through AIG models, which is linked to the idea of optimizing writing, producing agile writing, generating research suggestions adapted to the topic under study, saving time, creativity, and efficiency - words and sayings wrapped up in the network of discourses on AIG⁹ - has contributed to strengthening a representation of the subject-author who, under the illusion of being and standing at the origin of his words, is solely responsible for them.

In this regard, as Stiegler (2018) and Auroux (1998), for example, warn us, all these writing technologies transform the way we consume and interact with information, and, we might add, with the production of knowledge. The importance of technologies, whatever they may be, in human society is undeniable, since, ultimately, transformations in a society are constituted by the invention/creation of new instruments and their integration with complex human behaviors and relationships (Auroux, 1998, p. 321). This perception, therefore, transcends a pragmatic view of the functionality/applicability of a technological tool, of its neutrality, as a mere support with which a task can be performed. This understanding of technology, as in the case in question, AIG, leads us to take it at its historical-discursive depth, that is, to consider that its functioning (and functionality) are implied as a condition for the production of processes of subjectivation, as well as part of the modes of existence of the subject (Dias, 2018) and the objects generated by it, in our contemporary world. In this sense, the impact of AIG cannot be underestimated, especially when it comes to its intervention in the modes of formulation and circulation of knowledge, making the text generated by it work as a digital script/textuality whose materiality is not reduced to what is shown on the textual surface, nor is it reflected in the creation of texts that are syntactically well-structured and imbued with semantic coherence.

Furthermore, we refer back to Orlandi, who has long warned us that language technologies, such as writing, seem to render the notion of text obsolete.

Contrary to common sense, which imagines that we have now finally freed ourselves from text, the very notion of text must be the focus of our attention, as it is the basis for all these developments. And for new language technologies, which are, in turn, new/other writing technologies. Knowing how to construct a text means knowing how to theoretically construct the new formulations that are produced today. New forms of textuality. In new contexts of realization, that is, as it is said in Frenchdiscourse analysis, [in] new conditions of production (Orlandi, 2001, p. 15).¹⁰

1 Scientific Writing, its Text; The Author-Function, its Reader-Effect

Writing, the social practice from which all texts originate, from a discursive perspective- whether fictional or not- is a place of interpretation, formed in a symbolic space where language and discourse coexist simultaneously (Correa, 2002). In other words: in the activity of writing, the subject recovers what has already been said, indicating its insertion into interdiscourse and repeatability. The practice of writing therefore works with discursive memory, with meanings already inscribed, which are recovered and/or remembered, and with forgetting, since it is necessary to forget pre-existing meanings in order to be able to speak (Indursky, 2016). In other words, such forgetting results in the subject’s illusion of being the origin of meanings and in the impression of the reality of thought.

On the contrary, writing through AIG operates with a linearized metallic memory (Orlandi, 1996);¹¹ that reduces discursive knowledge (what has already been said) to a package of ideologically equivalent information, the effect of which is completeness and transparency. It is a memory that works with the quantity of information, with movements of repetition, reproduction, and replication. The author goes on to argue:

Its distinctive feature is that it is horizontal (rather than vertical, as defined by Courtine), meaning that there is no stratification in its process, but rather serial distribution, in the form of addition and accumulation: what has been said here and there and elsewhere comes together as if forming a network of affiliation rather than just a sum. Quantity rather than historicity (Orlandi, 2001, p. 5).¹²

As for the epistemic author-position,¹³ it is undoubtedly a scientific demand, made possible in and by scientific discursive practices, which goes beyond a richer vocabulary, a more articulate syntax, and more complex texts- generated by AIG- as its creators would have us believe. None of this corresponds to the reality of language and (scientific) discourse in the production of meaning (Correa, 2022), as we aim to discuss below, from the perspective of French-oriented discourse analysis.

Having said that, we intend to address the concepts of scientific writing and its correlate, scientific text, the epistemic author function, and also its correlate, the reader-effect, which have been guiding practices of this type of writing through the generation of texts by AIG, based on prompts¹⁴ (for example, “Write an introduction to an article discussing the effects of fake news”¹⁵).

Following the repeatability of the concepts of writing, text, author function and its double- the reader-effect- and epistemic subject, we operationalize them through an analytical reading of a text generated by ChatGPT, guided by the following prompt: define the concept of fake news and explain its importance for French-oriented discourse analysis, providing citations and references. Reiterating our position, the notion of writing mobilized here, in light of French-discourse analysis, is conceived in an unavoidable relationship between social practice and process, which implies admitting that writing is anchored in history, in in society, in ideology, constituting itself as a symbolic space, a place of interpretation, in a work of memory and construction of subjectivities (Orlandi, 2001).

Thus, as a social practice, “writing is one of the ways in which individuals relate to history, their time, society, in short, to inscribe themselves in the social body” (Indursky, 2006, p. 35).¹⁶ From this perspective, with regard to scientific writing, we consider that, in a literate society such as ours, this writing is one of the necessary conditions for the subject to insert themselves into the discursive practices of the field of knowledge that is taken as a reference to theoretically and methodologically situate their object of study, and thus politically and epistemologically mark their belonging to this field. In this direction, the act of writing, even if done alone, is subject to the injunction, at the same time, of the institutional space to which the subject is linked and the disciplinary field that governs the functioning of such writing practice and, to that extent, as a consequence, the sayings produced and put into circulation in and by the field of knowledge in question.

In short, writing is a form of subjectification. Thus, we understand scientific writing as linked to the discursive orders of the scientific institution, which gives body and materiality to different forms of knowledge, produced and circulated in the sphere of research activities. And it is this discursive work of writing that inscribes the subject both in the discursive networks of memory and in the authorial function, we might add, in epistemic authorship (Indursky, 2016).

As a process, writing “consists of the unfolding of a work that begins in the intersubjective encounter, producing any linguistic fact, and which, at least indicatively, is marked in the thread of discourse” (Corrêa, 2013, p. 89).¹⁷ The concept of process, linked to that of work, encompasses the notion of “writing as a continuous process, an activity that does not end, which, even though it can be suspended, paused, or interrupted, has no end,” as pointed out by Indursky (2016, p. 32).¹⁸ According to the researcher, conceiving writing as a process means that it can be continuously resumed and continued, transformed, mobilizing other and/or new discursive threads. And, by submitting a text to the production of interdiscourse - that which has already been said - “writing becomes an effort to intertwine discursive threads from intertext and/or interdiscourse” (Indursky, 2016, p. 35).¹⁹ We highlight here the word ‘effort’: contrary to this arduous work of the subject in the practice of writing, the discourse on the use of AIG for the generation of scientific text asserts statements such as “simplification for the production of academic texts,” “optimization of the research process,” among others, all linked to an ideology of the global world of facilitation, of the “convenience store.”²⁰

In line with what has been said and echoing the position of our reflection, in this text we consider the issue of scientific writing within the framework of discourse production regulation described by Foucault (1981).²¹ In other words, the functioning of discursive practices related to scientific writing, in the institutions that govern them, are structured in a system linked to the organization, qualification, and distribution of knowledge generated in this field. Under Foucault’s perspective (1981),²² this implies that entry into the domain of scientific writing discourses has historically been subject to mechanisms of control and power that determine practices and regulate the subjects who aspire to integrate them. One of these control mechanisms is precisely authorship, as discussed in The Order of Discourse²³ by the philosopher. In this work, Foucault argues that every society imposes an order on discourse, distancing it from incessant noise or uncontrolled murmuring. “The image of the absolutely free word, available to all, circulating without any limitation, rather than a dream, tends to be experienced in a society as a nightmare: the threat of falling into a kind of ‘discursive savagery’” (Alves, 2015, p. 79).²⁴ We are experiencing the manifestation of such fear at a time when the order of scientific discourse is being shaken and its ordering procedures are being challenged by new writing technologies that use autonomous systems to generate scientific texts.

From this perspective, we believe we can extend these assumptions about scientific writing practice to the questions we raise about its new practices through Generative AI, because, as we seek to discuss in the next section, writing technologies and their strategies via prompts, mobilized by subjects²⁵ in their commercial videos, seem to be anchored to the idea that scientific writing is an activity marked by the objectivity and neutrality of those who practice it, thus nullifying the subject’s inscription in the order of scientific discourse through the control of meaning production. As a result, procedures and techniques are suggested that are exercised on the processes of text formulation, “the construction of its object(s), the forms of its enunciation, and the subjects that are inscribed therein” (Rodrigues; Silva, 2019, p. 51).²⁶

Returning to the text, the verbal materiality of the practice of writing, the place where interpretation is textualized, is formed in a symbolic space where memory and forgetting work together. That is, the subject, in his text, recovers knowledge, which points to his inscription in interdiscourse and in the order of repeatability, as already mentioned. Indursky (2016) states that a text is a provisionally structured unit that can be deconstructed at any moment under the effect of the reading work of a reader-subject. Thus, the subject-author has

the possibility [...] of occupying the discursive place of the other, of the reader, returning to one’s own text to make a destabilizing reading of its effect of completeness, either to self-correct or to bring in other discursive threads that, for various reasons, could not be contemplated previously and weave them with those that were already there (Indursky, 2016, p. 31).²⁷

‘Weaving’ the discursive threads is a task of reading the author function, which seeks, in the materiality of the text, the effect of completeness and finish by means of textualization processes. In other words, the subject, in the author function, in the formulation of the text, invests in a work with meanings, giving them shape in the text, as Orlandi (2004) reminds us. In this work of textualization, resulting from a gesture of interpretation by the subject, all of their discourse, “updated” in the thread of discourse (intradiscourse, in the linearization of discourse), is constitutively traversed in and by interdiscourse (discursive memory). As a result, it is understood that the practice of writing results in a text that is woven from quotations - relationships with the intertext - from glosses, discursive paraphrases, and iterations. Such modes of intertextuality and interdiscursivity are primary characteristics of a scientific text. In other words, a scientific text is always an already inhabited object, in which its relationships with other texts and other discourses are made explicit.

In order to understand the author function, Foucault (1977) states we must take into account four characteristic features:

The author’s name is not a function of a man’s civil status, nor is it fictional; it is situated in the breach, among the discontinuities, which gives rise to new groups of discourse and their singular mode of existence. Consequently, we can say that in our culture, the name of an author is a variable that accompanies only certain texts to the exclusion of others: a private letter may have a signatory, but it does not have an author; a contract can have an underwriter, but not an author; and, similarly, an anonymous poster attached to a wall may have a writer, but he cannot be an author. In this sense, the function of an author is to characterize the existence, circulation, and operation of certain discourses within a society (Foucault, 1977, p. 124).²⁸

From the descriptions presented by the philosopher, we understand that the author function is determined by discursive positions, which, in scientific discourse, demand an epistemic position.

In The Archaeology of Knowledge & The Discourse on Language,²⁹ the philosopher argues that the author “is not identical with the subject of the statement,” but “in so far as it is an empty function” (Foucault, 1972, pp. 92-3).³⁰ This principle regarding the place of the author draws on the reasoning of a mathematical theory, according to which for each value of x, we have a value of f(x), where x is an empty place. Thus, to “find” the value of x, it is necessary to ask: “under what conditions and through what forms can an entity like the subject appear in the order of discourse; what position does it occupy; what functions does it exhibit; and what rules does it follow in each type of discourse?” (Foucault, 1977, p. 138).³¹ In this context, the notion of author is understood as an ordering principle that characterizes a certain way of being in discourse, which gives the subject a certain position. The position of the subject, in the author function of scientific discourse, is to activate in the discursive memory what has already been said in their field, in their theoretical framework, which allows them to attribute meanings to a given signifying materiality.

Furthermore, according to Orlandi (2001, p. 91), questioning Foucault’s theories, “the author function is performed in relation to the constitution of a place of interpretation defined by the relationship with the Other (interdiscourse) and the other (interlocutor).”³² The author is produced by the possibility of a gesture of interpretation that corresponds to him and that comes ‘from the outside,’ its constitution being possible only if what he produces is interpretable. Such interpretation requires organizing the dispersion of meanings in the linearity of speech - intradiscourse - producing new formulations, under the effect of unity resulting from its identification with a discursive formation, which dictates what can, what should, and how it can be said. Thus, for the author function to be constituted, its double - the reader-effect - must also be constituted (Orlandi, 1996; 2001). In this discursive interplay between author and reader, the reader’s act of interpretation is not limited to verifying meanings, as if there were a meaning behind the text, but involves exposing oneself to its opacity, which, from a discursive perspective, presupposes that meanings may be different, including those that are unforeseen. This concept of interpretation, focusing on the author, places them in a position that historicizes their discourse; that is, it inscribes discourse within the repeatable, within the interdiscourse, within what Foucault calls a ‘node in a network’ (Foucault, 1972).³³ In other words, and in the words of Orlandi (2001, p. 70), “the subject only becomes an author if what he produces is interpretable.”³⁴ Interpreting, therefore, is “a work of meaning on meaning” (Pêcheux, 2006, p. 51).³⁵

In this context, what repetition are we referring to? Not the formal repetition that, in scientific writing, is carried out by the injunction of Associação Brasileira de Normas Técnicas [Brazilian Association of Technical Standards] - ABNT standards or other normative techniques. That is, scientific standards for referencing other texts only mention a prescriptive description of the other’s discourse, considering its forms- direct quotation, indirect quotation, and quotation of a quotation. This point of view on the representation of another’s discourse suggests very little of what the act of quoting (the repeatable) actually means, given the limits of the functioning of technical discourse.³⁶

We refer to historical repetition, in which interpretation occurs, since the same/what has already been said are part of discursive memory, which enables the subject to formulate and construct their statement within the possible repetitions. In other words, from a discursive perspective, the author produces other meanings, introducing interdiscourse into other statements formulated by them. But

[a] statement exists outside any possibility of reappearing; and the relation that it possesses with what it states is not identical with a group of rules of use. It is a very special relation: and if in these conditions an identical formulation reappears, with the same words, substantially the same names - in fact, exactly the same sentence - it is not necessarily the same statement. (Foucault, 1972, p. 89).³⁷

What is stated in this principle of the relationship between sentence and utterance is that the utterance is not reduced to a linguistic unit, “superior to the phenomenon and the word, inferior to the text,” as the philosopher states (Foucault, 1972, p. 106).³⁸ And having clarified that the utterance is also a function, it is worth addressing it in relation to the function of authorship.

I was dealing with enunciative function that involved various units (these may sometimes be sentences, sometimes propositions; but they are sometimes made up of fragments of sentences, series or tables of signs, a set of propositions or equivalent formulations); and, instead of giving a ‘meaning’ to these units, this function relates them to a field of objects; instead of providing them with a subject, it opens up for them a number of possible subjective positions; instead of fixing their limits, it places them in a domain of coordination and coexistence; instead of determining their identity, it places them in a space in which they are used and repeated. (Foucault, 1972, p. 106).

To resume, “the reader-effect represents, for the author function, its constitutive exteriority - memory of discourse, historical repetition” (Orlandi, 2001, pp. 74-75).³⁹ “It is in the eyes of the reader that the inscriptions arrive, that the fragments appear reunited, that nothing is lost and everything is done (in the provisional unity of the text),” given that in him (the reader) there is dispersion [...]; “it is in the destination of a text that its unity is found, not in its origin” (Furlanetto, 2015, p. 159).⁴⁰

In Semântica e Discurso [Semantics and Discourse], Pêcheux develops the thesis of subjective illusion in the “imaginary” space of enunciation. He perceives in this subjective illusion a double process: one of imposition and one of dissimulation. The subject is constituted by the unconscious (the Other) and ideology (the historical), which give it a place and at the same time conceal the illusion of autonomy resulting from this attribution (thesis of interpellation, subject-effect). It is from this perspective that the subject form (Pêcheux’s reading of Althusser) is created within a given discursive formation, accounting for the imaginary function of subjectivity. Once individuals are interpellated as subjects of discourse, they identify with the subject form (historical subject) of discursive formations, which regulate what can and should be said. In the case of this study, individuals are interpellated as subject-researchers within the scientific discursive formation, which regulates the forms and modes of saying.

Pêcheux (1995), in the same work, reviews Althusser’s postulates on Ideology and ideological state apparatuses and goes on to consider the impasse of an “eternal” ideology, which is not identical to itself (tension between unity/difference, same/other, heterogeneity). The discursive formation in which the subject is inserted, therefore, reflects this tension. This results in its constant reconfiguration: words, expressions, and statements change; losses, substitutions, additions, and transfers occur constantly within a given discursive formation. Thus, there may be several subject-positions in each discursive formation- identification, disidentification, and counteridentification. The positions of the researcher/epistemic subject in scientific text, according to the order of scientific discourse, are those of refutation and/or criticism- through disidentification and/or counteridentification with a given epistemology; and identification, but not through mere repetition, but rather through the gesture of interpretation that enables, in turn, the formulation and reformulation of other enunciations.

2 Proposing an Analytical Reading

In order to operationalize the categories discussed above, we set out to exercise the “writing” of a text by ChatGPT,⁴¹ as already announced, based on the prompt define the concept of fake news and explain its importance for French-oriented discourse analysis, indicating citations and references, guided by the hypothesis that the generated text could compose, perhaps, the justification for the choice of a theoretical-conceptual framework for the study of fake news in a scientific text such as an article, term paper, dissertation, or thesis. And, to that extent, the choice of the theme, fake news, is not arbitrary, considering the fact that it is the subject of discussions and research in different disciplinary fields. Regarding the choice of the disciplinary field, French discourse analysis, it is important to mention that this is not explained by our affiliation with it, but rather by its broad and systematic investment in themes and problems that encompass the practices of disinformation in contemporary society, which, however, is beyond the scope of this study.

To observe how the text generated by this writing technology works, we extracted sequences created by the chatbot in response to our requests, as summarized in Figure 1. It is important to note that there was no interaction between us and the versions of ChatGPT used, so there was no user profile history to interfere with the results generated by the prompt. The criteria for organizing the table into three columns takes into account the command/prompt, some discursive sequences, which attempt to define the discursive fact on screen and explain its importance for French-oriented discourse analysis, as well as indicate the sources and references used by the tool, given that the main concern of those who sell manuals for scientific writing via AIG seems to be the absence of plagiarism.

Figure 1
Chat interaction⁴⁵ description and the text composition⁴⁶

The following are the discursive sequences taken as the object of analysis.

Regarding the prompt in question, from the discursive perspective adopted here, it can be understood as one of the factors that make up the conditions for the production of the text generated by this AIG. This is because the content is created (to use the terminology of this technology) under the injunction of a command/instruction, which informs the language model of the AIG to perform a task, which follows a given request to determine the structure and content of the text to be generated.

Taking this perspective, as Pêcheux (1969) asserts, requires transposing the immediate situation of the prompt’s formulation as a mere linguistic expression. We thus seek to understand it in relation to other sayings and utterances from other places that traverse and constitute the discursivity of scientific writing practices- and, equally, in relation to the subject positions occupied by individuals each time they take a position within this discursive practice. As observed in our study, AI guides, mobilizing what is known as a prompt library, provide a collection of predefined commands that serve as models to streamline the creation of prompts and support researchers in their scientific tasks.

Based on this collection, the subject, relying on schematic models, can adapt/reformulate these prompts to meet the specific writing needs of their text,⁴³ which, however, correspond to organizational standards and requirements of what is conceived as scientific writing and a methodological direction based on a positivist view of scientific work (Rodrigues; Silva, 2019). Given this possibility, we set out to ‘practice’ the writing of a text by ChatGPT by drawing on a suggested prompt, formatted in a section of an e book,⁴⁴ limited to commands that address issues related to the production of theoretical foundations, the disciplinary field, and other aspects of scientific work. Returning to the prompt, it is formulated as follows: Define [concept/theory] and explain its importance to the field of [field of study], and then we added, indicating citations and references.

By including citations and references in this request, we did so based on the assumption that scientific texts are already inhabited by other sayings, by things already said, in which references to scientific authority are required. In other words, it is part of scientific discourse to cite other texts, to engage in dialogue with their authors, thus “explaining the intertextuality that underpins their formulations and recognizing the different author functions that intervene throughout the text, recognition guaranteed by the citations” (Orlandi, 2008, n.p).⁴⁵ As already discussed, citing requires the subject to take a position in relation to those they cite and, by extension, to the disciplinary field in which that subject is enrolled and, fundamentally, to the scientific discursive formation that regulates the forms and modes of their discourse, which, in turn, inscribes the subject in a discursive position.

Let us also focus on the verbal forms “define” the concept of “x” and “explain” the importance of “x,” from a given theoretical-conceptual perspective of a given disciplinary field, present in the structure of this prompt. They can be understood discursively as ways of saying that, in the order of scientific discourse, they bring into play discursive meanings and positions that presuppose, on the part of the subject of this and in this writing, the appropriation of theoretical-conceptual and methodological knowledge of the disciplinary field in question. Thus, defining something and explaining it are discursive operations that, in the practice of scientific writing, are intertwined, insofar as the activity of explaining presupposes that of defining, of making known, mobilizing, for this purpose, in the formulation of the text, ways of saying through which one seeks to explain, comment on, substantiate, and demarcate, for example, the object under study. The effect of meaning of these discursive operations contributes to demonstrating the theoretical and methodological rigor and reliability of the research.

Thus, with the work of appropriation, which involves an act of interpretation, as Pêcheux (1997) and Orlandi (2004) remind us, the subject, in their writing, participates in the collective construction of the knowledge that constitutes the discursive (theoretical) memory of the disciplinary field. In this way, when revisited, this discursive memory is reconstituted and reformulated in writing, manifesting itself in the textualization of the work, through writing, and materializing the affiliation to the field by the researcher who inscribes themselves in it.

Furthering this analysis, we selected the French-oriented discourse analysis sequence, which covers influences from thinkers such as Michel Foucault, Jacques Derrida, and Pierre Bourdieu, exploring how discourse constructs and reflects power relations in society. For the analysis of fake news, these theorists offer essential tools [...] in which there is a predominance of a formulation that is evident, in a consensual way, as if, for the area in question, the theorists cited were read in the same way by their peers, without the work of scientific debate. Thus, the subject-position is projected in this writing based on formulations that are presented as unquestionable truths. References to these theorists indicate how, in writing technologies, the information and/or content generated by AIG, under algorithmic logic, is disconnected from any historical or cultural memory, in this case, the theoretical memory of the French DA field, as discussed above. Here, mythical understanding reaches its fullness by being expressed in a universal logical-mathematical language, which, by its nature, lacks memory (Pêcheux, 1997). In other words, metallic memory, under the order of producing by quantity, of circulating the same, loses the historicity of meaning, because by bringing in authors who do not belong to the field of AD, it places them in the same epistemological framework, allowing us to see that the author function it occupies is not affiliated with the epistemological discursive formation in question, which dictates what can and should be said, but also who can and should be quoted.

In the sequence For Michel Foucault, truth is not a universal essence, but rather a social construct that is articulated through institutional discourses and practices. In his work on the “history of truth,” Foucault argues that what is considered true in a society is the product of a constant struggle for power and control over discourse; if we take into account that the prompt requested an explanation of the importance of studying fake news for the analysis of French-oriented discourse, we can see that such an explanation is not present in the formulation, as the relationship between the notion of truth postulated by Foucault and fake news is not linguistically marked. As Orlandi (2004, p. 14) points out, reformulation is not simply “the explicitation of what was already there.” Or, as Courtine (2014, p. 201. Author’s emphasis)⁴⁶ describes it,

each formulation, whose repeatability is based on a statement, must be considered in a possible reformulation, that is, it is only a “supporting point” for access to interdiscourse, and is therefore not transparent, capable of being used in any text.⁴⁷

That reminds us of “phrases without text,” whose process Maingueneau (2014, p. 14) calls detachability, that is, “statements that are autonomous from a textual point of view (there is no need to consider what precedes and what follows to understand them) and from an enunciative point of view (they are generalizations).”

The same interpretation can be proposed by referring to Derrida: Language is a game of forces and meanings that never completely stabilize (Derrida, 1967, p. 77). [...] This idea is relevant for understanding how fake news can be understood. For the field of French-discourse analyses, such a concept of language may be consensual, but for those who propose to enter the realm of scientific discourse, specifically in FD, which conceives the functioning of language from a discursive point of view, it is necessary to argue about this principle of the area, explaining its functioning based on the reading of the object of study in question. But machine memory does not offer this explanation because it mobilizes meanings that are not accessible to it, because it does not have them, given that the algorithm has an erroneous conception of language and knowledge. Our reader may argue: but these paragraphs can only be used as a starting point for reasoning about the relationship between fake news and French-discourse analyses. If so, this statement shows us how the work of the reader-effect of the author-function is necessary, an idea not propagated by sellers of AIG user manuals for scientific writing.

Finally, in the text generated by ChatGPT, it can be seen that the statements are adequately written from the point of view of syntax and scientific standards (APA) with regard to the use of the sources consulted, which, in turn, are textualized via indirect and direct citation. However, as Orlandi (2004, p. 15) reminds us, the text cannot develop in any direction, and this can only be understood “if we do not think about the text in terms of its organization, but rather in terms of its signifying order.”⁴⁸ To be a text, “it is not enough to be written and more or less adequate to the rules of the language. It needs to have its own marks from the person who wrote it” (Barzotto, 2016, p. 144).⁴⁹ The question that arises is: do novice researchers know this? This is a conversation for another time, given the contingent limitations of this text.

Concluding Remarks

In view of the principles that guide the concept of author function, we argue that the relationship between subject, discourse, and power established by Foucault constitutes a possible lens for reading this category in the practices of scientific text production by autonomous systems. In Archeology of Knowledge & The Discourse on Language,⁵⁰ as already mentioned, Foucault argues that the author “is not identical with the subject of the statement,” but “in so far as it is an empty function” (Foucault, 1972, pp. 92-3).⁵¹ It is precisely from this concept of function that we would like to propose a possible interpretation of the illustrative reading undertaken here.

Let us return to Foucault’s questions about the author function: “under what conditions and through what forms can an entity like the subject appear in the order of discourse; what position does it occupy; what functions does it exhibit; and what rules does it follow in each type of discourse?” (Foucault, 1977, p. 138).⁵²

The provisional answer we can offer is that we understand that there is an inverse function f^-1(x), contrary to that predicted by Foucault - f(x) - of authorship functioning in writing practices via AIG. This is because the construction of authorship provides for formulation within the formulable. That is, the author produces a place of interpretation that occurs through discursive memory and the subject-position assumed within a given FD, which regulates what can and should be said. Their work of reading and producing texts is not reduced to imitation (formal repetition) and copying. If this were the case, it would prevent the possibilities of constructing subjectivity. Contrary to this postulate, the inverse function - f^-1(x) - of the construction of authorship through writing technologies occurs through a gesture of interpretation outside history, which therefore constitutes itself as a formula (and not as a formulation), as a rearrangement and not a re-signification (Orlandi, 2004), but “this does not mean that there is no production of authorship” (Orlandi, 2004, p. 17).⁵³ It occurs, but it is of a different nature, precisely because memory is different- it is metallic. As we know, from a discursive point of view, different materialities determine different processes of signification (Orlandi, 2004). Metallic memory linearizes⁵⁴ interdiscourse, reducing discursive knowledge to a package of information. The functioning of the human mind, on the other hand, differs fundamentally from ChatGPT and similar models, which operate as robust statistical machines, analyzing enormous volumes of data to identify patterns and determine the most likely answers to scientific questions, among others. In contrast, the human mind is an incredibly efficient and sophisticated system, capable of working with limited amounts of information, not reducing itself to identifying mere correlations between data, but formulating meaningful explanations, which we do not identify in the sequences under analysis.

As Foucault argued, there are different historical forms of the author function, because the author’s relationship with writing is never the same throughout the history of societies. In this, it seems to us, this relationship is constructed by a technology of writing whose reader subject-position engenders a meaning that is constructed by another - algorithms - and not by the researcher subject-position, which questions, reads, researches...

Along with Alves (2015, p. 81. Author’s emphasis), we understand that

[the] author, taken as a modern construct, is undergoing a profound transformation. We are facing a radical reformulation of the “order of discourses,” within which we see the emergence of new technologies of power and subject-positions, which seem to point to the “anonymity of the murmur” or to unprecedented forms of unification and appropriation of discourses, now taken as information in a large, shared, and changing network.⁵⁵

The most “dangerous” thing, recalling Foucault’s statement and returning to the question posed in the title of this text, is that neoliberal discourse has sought to produce an empty subject-position, in the sense of a subject of science who only repeats- whether they are qualified or a beginner.

• 1
  The question that forms the title of this article is a quote taken from Foucault’s The Order of Discourse. In this work, reflecting on the internal procedures of discourse control, on what could lead to chance, to danger, the philosopher discusses how “In every society the production of discourse is at once controlled, selected, organized and redistributed by a certain number of procedures whose role is to ward off its powers and dangers, to gain mastery over its chance events, to evade its ponderous, formidable materiality.” (Foucault, 2004, pp. 8-9). Ironically, he asks: where, after all, is the danger? (Foucault, 2004, pp. 8-9).
• 1
  FOUCAULT, Michel. What Is an Author? In: Language, Counter-Memory, Practice. Selected Essays and Interviews. Ithaca: Cornell University Press, 1977, p. 138.
• 2
  FOUCAULT, Michel. The Order of Discourse. In: Untying the Text. A Post-Structuralist Reader, London and Henley, 1981, p. 138.
• 3
  According to Lana (2023), there is no consensus on the definition of Artificial Intelligence. The first reference to AIG was made at the Dartmouth Conference in 1956 - a field of study and a set of technologies that employ software and physical devices to reproduce advanced capabilities typical of human intelligence (McCarthy et al. apudLana, 2023).
• 4
  In the field of language studies, research has been undertaken in many ways that focuses on the discursive functioning of Gen-AI and its effects on the relationship between subject and meaning. We cite here Gallo (2023), who discusses the status of the author in the face of new conditions of production in digital materiality, and Fernandes (2024), who problematizes the concept of authorship based on an analysis of texts produced by AIG in comparison with texts produced by students in school assignments.
• 5
  According to the ranking of a survey conducted by Semrush, a company specializing in digital marketing, Brazil ranks fourth among the countries that consume the most AIG (https://shre.ink/b1wj), behind only Indonesia, India, and the United States.
• 6
  In Portuguese: “instrução textual fornecida a um modelo de linguagem (LLM) para solicitar respostas ou gerar conteúdo específico.”
• 7
  In Portuguese: “re-organização da vida intelectual, re-distribuição dos lugares de interpretação, promovendo um deslocamento no funcionamento da autoria e da própria concepção de texto.”
• 8
  Critics would say that there has been a decline in reading and writing skills (brain rot); on the other hand, it is undeniable that there has also been progress in familiarization with digital technologies (knowledge of tools, word processing and image processing, remixing, etc.), at least among the portion of the population that has access to them. And, in this context, there is an increasing number of manuals being sold with suggestions for research writing prompts - articles, theses, dissertations, projects, final course assignments - offered by people who claim to be experts on the subject, given their doctoral training, regardless of their area of expertise (See LIN, Michael Pin-Chuan; CHANG, Daniel. Enhancing Post-Secondary Writers’ Writing Skills with a Chatbot: A Mixed-Method Classroom Study. Educational Technology & Society, v. 23, n. 1, 2020, pp. 78-92. Available at: https://www.jstor.org/stable/26915408. Access on: Feb. 20, 2025.)
• 9
  See, for example, https://jenni.ai/pt/for-researchers.
• 10
  In Portuguese: “Ao contrário do senso comum, que imagina que agora, enfim, nos libertamos do texto, a noção de texto deve ser objeto de toda nossa atenção, pois é ela que está na base de todos esses desenvolvimentos. E das novas tecnologias de linguagem que são, por sua vez, novas/outras tecnologias da escrita. Saber como se elabora um texto é saber elaborar teoricamente as novas formulações que hoje se produzem. Novas formas de textualidade. Em novos contextos de realização, ou seja, como se diz em AD, [em] novas condições de produção.”
• 11
  Also referred to as technical/artificial Stiegler (2018), or even a dead memory (Robin, 2016).
• 12
  In Portuguese: “A sua particularidade é ser horizontal (e não vertical, como a define Courtine), não havendo assim estratificação em seu processo, mas distribuição em série, na forma de adição, acúmulo: o que foi dito aqui e ali e mais além vai-se juntando como se formasse uma rede de filiação e não apenas uma soma. Quantidade e não historicidade.”
• 13
  Furlanetto (2003), when discussing subjectivity from the perspective of French-discourse analysis, refers to the concept of the epistemic subject to discuss the scientist’s self in research writing. Based on this, we then use the concept of epistemic author-position, understanding that the scientist’s self is challenged by the subject form of science.
• 14
  In Scalco’s manual, from which we generated the prompt for the illustrative reading of this text, the following examples of prompts for research writing can be found: “list the most influential authors and/or researchers in the field of [topic]; examine citation patterns of the [number] most cited articles on [topic]; compile a bibliography of important articles on the effects of air pollution on childhood asthma” [In Portuguese: [liste os autores e/ou pesquisadores mais influentes na área de [tema]; examine padrões de citação dos [número] artigos mais citados sobre [tema]; compile uma bibliografia de artigos importantes sobre os efeitos da poluição do ar na asma infantile.] (Scalco, n.p. pp. 23-29).
• 15
  In Portuguese: “Escreva uma introdução de um artigo que discuta os efeitos das fake News.”
• 16
  In Portuguese: “a escrita é um dos modos de que o sujeito lança mão para relacionar-se com a história, com seu tempo, com a sociedade, em suma, para inscrever-se no corpo social.”
• 17
  In Portuguese: “consiste no desenrolar de um trabalho que começa no encontro intersubjetivo, produtor de qualquer fato de linguagem, e que, ao menos indiciariamente, se marca no fio do discurso.”
• 18
  In Portuguese: “escrita como um processo contínuo, uma atividade que não se encerra, que, mesmo podendo ser suspensa, pausada ou interrompida, não possui um fim.”
• 19
  In Portuguese: “a escrita se torna um esforço para entrelaçar fios discursivos provenientes do intertexto e/ou do interdiscurso.”
• 20
  In Portuguese: “simplificação para a produção de textos acadêmicos”; “otimização do processo de pesquisa”; “loja de conveniência.”
• 21
  See footnote 3.
• 22
  See footnote 3.
• 23
  See footnote 3.
• 24
  In Portuguese: “A imagem da palavra absolutamente livre, ao alcance de todos, que circula sem qualquer limitação, ao invés de um sonho, tende a ser vivenciada em uma sociedade como um pesadelo: a ameaça de cairmos em uma espécie de ‘selvageria discursiva.’”
• 25
  The subjects we refer to are, for the most part, doctors with degrees in various fields of knowledge.
• 26
  In Portuguese: “a construção de seu(s) objeto(s), as formas de sua enunciação e os sujeitos que aí se inscrevem.”
• 27
  In Portuguese: “a possibilidade [...] de ocupar o lugar discursivo do outro, do leitor, retornando a seu próprio texto para sobre ele fazer uma leitura desestabilizadora de seu efeito de completude, seja para se autocorrigir, seja para trazer outros fios discursivos que, por diversos motivos, não puderam ser contemplados anteriormente e tramá-los àqueles que aí já se encontravam.”
• 28
  Footnote 2.
• 29
  FOUCAULT, Michel. Archeology of Knowledge & The Discourse on Language. New York: Pantheon Books, 1972.
• 30
  Footnote 30.
• 31
  Footnote 2.
• 32
  In Portuguese: “a função-autor se faz na relação com a constituição de um lugar de interpretação definido pela relação com o Outro (interdiscurso) e o outro (interlocutor).”
• 33
  Footnote 30.
• 34
  In Portuguese: “o sujeito só se faz autor se o que ele produz for interpretável.”
• 35
  In Portuguese: “um trabalho do sentido sobre o sentido.”
• 36
  In the material analyzed below, the main argument for generating texts via AIG is the citation and reference to the citation as characteristics of authorship.
• 37
  Footnote 30.
• 38
  Footnote 30.
• 39
  In Portuguese: “o efeito-leitor representa, para a função-autor, sua exterioridade constitutiva -, memória do dizer, repetição histórica.”
• 40
  In Portuguese: “É aos olhos do leitor que as inscrições chegam, que os fragmentos aparecem reunidos, que nada se perde e tudo está feito (na unidade provisória do texto)”; “é no destino de um texto que se encontra sua unidade, não em sua origem.”
• 41
  The version used was the free one - https://chatgpt.com/?model=auto - based on the assumption that it would be the most widely used by researchers in training. We also requested the same task for the paid AI-PRO version, using the same prompt. We obtained different versions in terms of textual formulation, but from a discursive point of view, the answers given did not differ.
• 45
  The terminology used in the field is HMI - a system that connects people to machines and their data - human-machine interaction.
• 46
  Appended are the complete discursive sequences that make up the metallic memory area activated by AIG for the composition of the text. It is important to emphasize that the quotations shown in Figure 1 were generated by ChatGPT based on the prompt provided by the authors and constitute the set of discursive sequences taken as the object of analysis in this study.
• 42
  In Portuguese: “Solicitação dirigida ao ChatGPT; Sequências; Indicação da fonte, pelo sistema autor-data, introduzida pelos autores a partir das referências indicadas pelo ChatGPT; Defina o conceito de fake news e explique sua importância para a análise de discurso de orientação francesa, indicando citações e referências; O conceito de fake news refere-se a informações falsas ou enganosas que são disseminadas com o intuito de enganar, manipular ou influenciar a opinião pública, muitas vezes em contextos sensíveis como política, saúde ou eventos sociais [...]; Sem fonte; A análise de discurso de orientação francesa, que abrange influências de pensadores como Michel Foucault, Jacques Derrida e Pierre Bourdieu, explora como o discurso constrói e reflete as relações de poder na sociedade. Para a análise de fake news, esses teóricos oferecem ferramentas essenciais para entender como a manipulação de informações pode afetar a construção social da verdade, a formação de subjetividades e o controle das narrativas dominantes; Referência aos teóricos “escolhidos” pela IAG; Para Michel Foucault, a verdade não é uma essência universal, mas sim uma construção social que se articula através de discursos e práticas institucionais. Em seu trabalho sobre a “história da verdade”, Foucault argumenta que o que é considerado verdadeiro em uma sociedade é produto de uma luta constante por poder e controle sobre o discurso. Em “A Arqueologia do Saber”, ele explica: “O discurso não é apenas uma forma de comunicação, mas uma prática que produz realidade e estabelece limites sobre o que pode ser dito e acreditado” (Foucault, 1969, p. 10); Foucault, M. (1969). A Arqueologia do Saber. Rio de Janeiro: Forense Universitária; Jacques Derrida, com sua abordagem desconstrutivista, oferece outra lente crítica para a análise de fake news. Derrida, em obras como “A Escritura e a Diferença”, argumenta que o significado de um texto ou discurso nunca é fixo, mas está sujeito a interpretações múltiplas, desconstruções e reinterpretações. Ele afirma: “A linguagem é um jogo de forças e sentidos que nunca se estabilizam completamente” (Derrida, 1967, p. 77); Essa ideia é relevante para entender como as fake news podem ser compreendidas: elas não apenas transmitem informações erradas, mas alteram o próprio campo de significação, criando ambiguidade e incerteza. O discurso das fake news constantemente se depara com a desconstrução do que é verdadeiro, oferecendo um “texto” manipulável e instável; Derrida, J. (1967). A Escritura e a Diferença. São Paulo: Perspectiva.”
• 43
  Custom requests allow creators to guide the advanced language model (GPT) to act in a specific way or perform a specific task (see https://learn.microsoft.com/pt-br/ai-builder/create-a-custom-prompt). Several videos by authors of prompt manuals highlight this possibility of the Chatbot.
• 44
  The prompt was created based on guidance from Scalco (n.n.).
• 45
  In Portuguese: “a explicitação da intertextualidade que sustenta suas formulações e o reconhecimento das diferentes funções-autor, que intervêm ao longo do texto, reconhecimento garantido pelas citações.”
• 46
  In Portuguese: “a explicitação do que lá já estava.”
• 47
  In Portuguese: “cada formulação, cuja repetibilidade se sustenta em um enunciado, deve ser considerada em uma possível reformulação, ou seja, é apenas um “’ponto de apoio’” de um acesso ao interdiscurso, não se configurando, assim, como transparente, passível de estar a serviço de qualquer texto.”
• 48
  In Portuguese: “se não pensarmos o texto em sua organização, mas o texto em sua ordem significante.”
• 49
  In Portuguese: “it is not enough to be written and more or less adequate to the rules of the language. It needs to have its own marks from the person who wrote it.”
• 50
  Footnote 30.
• 51
  Footnote 30.
• 52
  Footnote 2.
• 53
  In Portuguese: “isso não quer dizer que não haja produção de autoria.”
• 54
  From a mathematical point of view, it is not correct to say that AIG, especially large models (LLMs), is “linear.” In reality, its operation is highly dependent on nonlinear mechanisms, with stochastic elements that make its behavior non-deterministic.
• 55
  In Portuguese: “o autor, tomado como um construto moderno, está sofrendo uma profunda transformação. Estamos diante de uma radical reformulação da “ordem dos discursos”, no seio da qual vemos emergir novas tecnologias de poder e posições-sujeito, que parecem apontar para o “anonimato do murmúrio” ou para formas inéditas de unificação e apropriação dos discursos, agora tomados como informações em uma grande rede compartilhada e mutante.”
• 56
  Footnote 30.
• 57
  Note: This quoting corresponds neither to any Foucault’s texts in Portuguese nor in English. According to the IA: “Essa formulação exata não é uma citação literal (palavra por palavra) com uma página fixa, mas sim uma paráfrase teórica ou resumo conceitual feito por pesquisadores para explicar a tese central de Michel Foucault” [This exact formulation is not a literal quoting (word for word), as a fixed page, but a theoretical paraprase ou conceptual gathering produced by researchers to explain Michel Foucault’s central thesis].
• Reviews
  Due to the commitment assumed by Bakhtiniana. Revista de Estudos do Discurso [Bakhtiniana. Journal of Discourse Studies] to Open Science, this journal only publishes reviews that have been authorized by all involved.

ChatGPT - Generated text

Prompt 1: Define the concept of fake news and explain its importance for French-oriented discourse analysis, citing quotes and references.

The concept of fake news refers to fabricated, distorted, or deliberately false information disseminated with the aim of manipulating public opinion, creating confusion, or achieving some kind of political, economic, or social advantage. This information is often conveyed through social media, news sites, and other digital platforms, where it spreads rapidly, often without verification or fact-checking. The term “fake news” has gained popularity especially in the context of recent political events, such as the 2016 U.S. elections and the misinformation surrounding the COVID-19 pandemic.

● The relevance of Fake news for French-oriented Discourse Analysis

French-oriented discourse analysis, which draws on influences from thinkers such as Michel Foucault, Jacques Derrida, and Pierre Bourdieu, explores how discourse constructs and reflects power relations in society. For the analysis of fake news, these theorists offer essential tools for understanding how the manipulation of information can affect the social construction of truth, the formation of subjectivities, and the control of dominant narratives.

1. Michel Foucault and the Production of Truth

For Michel Foucault, truth is not a universal essence, but rather a social construct that is articulated through institutional discourses and practices. In his work on the “history of truth,” Foucault argues that what is considered true in a society is the product of a constant struggle for power and control over discourse. In “The Archaeology of Knowledge & t

The Discourse on Language,”⁵⁶ he explains:

“Discourse is not only a form of communication, but a practice that produces reality and establishes limits on what can be said and believed” (Foucault, 1969, p. 10).*⁵⁷

In this context, fake news can be understood as a form of discursive manipulation that aims to reconfigure the truth, creating distorted or alternative versions of the facts that favor specific interests. The dissemination of fake news is a practice that disputes control over social narratives and, consequently, over the realities that people accept as true.

2. Jacques Derrida and the Deconstruction of Meaning

Jacques Derrida, with his deconstructivist approach, offers another critical lens for analyzing fake news. Derrida, in works such as “Writing and Difference,” argues that the meaning of a text or discourse is never fixed, but is subject to multiple interpretations, deconstructions, and reinterpretations. He states:

“Language is a play of forces and meanings that never completely stabilize” (Derrida, 1967, p. 77).

This idea is relevant to understanding how fake news can be understood: it not only conveys misinformation, but also alters the very field of meaning, creating ambiguity and uncertainty. Fake news discourse constantly confronts the deconstruction of what is true, offering a manipulable and unstable “text.”

3. Pierre Bourdieu and the Power of the Media Field

Pierre Bourdieu offers an interesting view on how fake news can be analyzed from his perspective of symbolic power. In “Distinction,” he states that the media field is a space where struggles for symbolic power take place, and the media plays a fundamental role in legitimizing certain discourses and marginalizing others. Bourdieu writes:

“The media has immense power to shape reality, as it is capable of creating a field of meanings that shapes what is possible to think and say in society” (Bourdieu, 1997, p. 120).

From this perspective, fake news functions as a tool for manipulating this symbolic field by disseminating distorted narratives that reinforce certain ideologies, altering the collective understanding of events, facts, and social problems.

● Conclusion

The analysis of fake news within the French tradition of discourse analysis reveals how these phenomena of disinformation are not just isolated “lies,” but discursive practices deeply rooted in power relations and strategies of symbolic control in society. Authors such as Foucault, Derrida, and Bourdieu help us understand that fake news is a field of dispute over truth, legitimacy, and control over opinion formation.

References:

Foucault, M. (1969). The Archaeology of Knowledge. Rio de Janeiro: Forense Universitária.

Derrida, J. (1967). Writing and Difference. São Paulo: Perspectiva.

Bourdieu, P. (1997). Distinction: A Social Critique of the Judgment of Taste. São Paulo: Edusp.

Research Data and Other Materials Availability

The contents underlying the research text are included in the manuscript.

References

• ALVES, Marco Antônio Sousa. A autoria em questão a partir de Foucault: autor, discurso, sujeito e poder. Matraga - Revista do Programa de Pós-Graduação em Letras da UERJ, [S. l.], v. 22, n. 37, 2015. Disponível em: https://www.e-publicacoes.uerj.br/index.php/matraga/article/view/19932 Acesso em: 9 mar. 2025.
  » https://www.e-publicacoes.uerj.br/index.php/matraga/article/view/19932
• AUROUX, Sylvain. A filosofia da linguagem Campinas, SP: Pontes Editores, 1998.
• BARZOTTO, Valdir, Heitor. Leitura, escrita e relação com o conhecimento Campinas, SP: Mercado de Letras, 2016.
• CARMO, Carlos Roberto Souza; CARMO, Renata de Oliveira Souza; MELO, Guilherme Diniz de. A inteligência artificial e os desafios na avaliação da escrita acadêmica. Cadernos da FUCAMP, v. 21 n. 53, pp. 1-25, 2022. Disponível em: https://revistas.fucamp.edu.br/index.php/cadernos/article/view/2919 Acesso em: 23 de fev. de 2025.
  » https://revistas.fucamp.edu.br/index.php/cadernos/article/view/2919
• CHARTIER, Roger. O algoritmo é o contrário do desejo. Entrevista concedida a Juan M. Zafra, publicada originalmente em The Conversation, em 17/09/2022 e reproduzida pela IHU - Revista do Instituto Humanitas Unissinos, on line, em 21 de setembro de 2022. Disponível em: https://www.ihu.unisinos.br/categorias/622297-o-algoritmo-e-o-contrario-do-desejo-entrevista-com-roger-chartier Acesso em setembro de 2025.
  » https://www.ihu.unisinos.br/categorias/622297-o-algoritmo-e-o-contrario-do-desejo-entrevista-com-roger-chartier
• CORRÊA, Manoel Luiz Gonçalvez. Bases teóricas para o ensino da escrita. Linguagem em (Dis)curso, Tubarão, SC, v. 13, n. 3, pp. 481-513, set./dez. 2013. Disponível em: https://www.scielo.br/j/ld/a/ndCmMjqNqdfb4CsYhkBvhQP/abstract/?lang=pt Acesso em 02 de dez. de 2025.
  » https://www.scielo.br/j/ld/a/ndCmMjqNqdfb4CsYhkBvhQP/abstract/?lang=pt
• CORRÊA, Manoel Luiz Gonçalvez. Letramentos em perspectiva histórica: do império da escrita aos sonhos do pós-pandemia. Muiraquitã: Revista de Letras e Humanidades, [S. l.], v. 1, n. 1, 2022. DOI: 10.29327/210932.1.1-2. Disponível em: https://periodicos.ufac.br/index.php/mui/article/view/5829 Acesso em: 23 fev. 2025.
  » https://doi.org/10.29327/210932.1.1-2.» https://periodicos.ufac.br/index.php/mui/article/view/5829
• COURTINE, Jean Jacques. Análise do discurso político: o discurso comunista endereçado aos cristãos. 1.ed. São Carlos/SP: Ed. UFSCar, 2014.
• DIAS, Cristiane. Análise do discurso digital: sujeito, espaço, memória e arquivo. Campinas, SP: Pontes Editores, 2018.
• FERNANDES, Carolina. A autoria em textos produzidos por inteligência artificial e por alunos em uma perspectiva discursiva. Revista da ABRALIN, [S. l.], v. 23, n. 2, pp. 214-235, 2024. Disponível em: https://revista.abralin.org/index.php/abralin/article/view/2183 Acesso em 02 de fev. de 2026.
  » https://revista.abralin.org/index.php/abralin/article/view/2183
• FOUCAULT, Michel. O que é um autor Pontinha - Portugal: Nova Vega, 1992.
• FOUCAULT, Michel. A ordem do discurso São Paulo: Edições Loyola, 2004.
• FOUCAULT, Michel. A arqueologia do saber Trad. Luiz Felipe Baeta Neves. 8 ed. Rio de Janeiro: Forense Universitária, 2014.
• FURLANETTO, Maria Marta. A morte do autor, a leitura de outrem e a construção autoral. In: FLORES, Giovanna. G. Benedetto.; NECKEL, Nádia. Régia. Maffi.; GALLO, Solange Maria Ledo. (Org.). Análise do discurso em rede: cultura e mídia. Campinas, SP: Pontes, 2015.
• FURLANETTO, Maria Marta. Sujeito epistêmico e materialidade do discurso: o efeito de singularidade. Linguagem em (Dis)curso, Tubarão, v. 3, Número Especial, pp. 91-119, 2003. Disponível em: https://portaldeperiodicos.animaeducacao.com.br/index.php/Linguagem_Discurso/article/view/248/0 Acesso em 06 de fev. de 2024.
  » https://portaldeperiodicos.animaeducacao.com.br/index.php/Linguagem_Discurso/article/view/248/0
• GALLO, Solange Maria Ledo. ChatGPT: hiperautor ou não autor? Traços de Linguagem, Cáceres, v. 7, n.1, pp. 84-95, 2023. Disponível em: https://periodicos.unemat.br/index.php/tracos/article/view/11199 Acesso em 02 de fev. de 2026.
  » https://periodicos.unemat.br/index.php/tracos/article/view/11199
• INDURSKY, Freda. As determinações da prática discursiva da escrita. Revista Desenredo, [S. l.], v. 12, n. 1, 2016. Disponível em: https://ojs.upf.br/index.php/rd/article/view/5954 Acesso em: 20 fev. 2025.
  » https://ojs.upf.br/index.php/rd/article/view/5954
• LANA, Pedro de Perdigão. A autoria das obras autonomamente geradas por inteligência artificial e o domínio público Coimbra: Edições Almeida, 2023.
• LIN, Michael Pin-Chuan; CHANG, Daniel. Enhancing Post-Secondary Writers' Writing Skills with a Chatbot: A Mixed-Method Classroom Study. Educational Technology & Society, v. 23, n. 1, 2020, pp. 78-92. Disponível em: https://www.jstor.org/stable/26915408 Acesso em: 20 fev. 2025.
  » https://www.jstor.org/stable/26915408
• MAINGUENEAU, Dominique. Frases sem texto Trad. Sírio Possenti et alii São Paulo: Parábola, 2014.
• ORLANDI, Eni Puccinelli. O que é linguística São Paulo: Brasiliense,1986.
• ORLANDI, Eni Puccinelli. A linguagem e seu funcionamento: as formas do discurso. Campinas, S.P.: Pontes; 1996.
• ORLANDI, Eni. Puccinelli. Discurso e texto: formulação e circulação dos sentidos. 2. ed. Campinas: Pontes, 2001.
• ORLANDI, Eni Puccinelli. Interpretação: autoria, leitura e efeitos do trabalho simbólico. São Paulo: Pontes, 2004.
• ORLANDI, Eni Puccinelli. Silêncios: presença e ausência. ComCiência, n. 101, n.p., Campinas, 2008. Disponível em: https://comciencia.scielo.br/scielo.php?script=sci_arttext&pid=S1519-76542008000400007 Acesso em: 23 de fev. 2025.
  » https://comciencia.scielo.br/scielo.php?script=sci_arttext&pid=S1519-76542008000400007
• PÊCHEUX, Michel. Semântica e discurso: uma crítica à afirmação do óbvio. Campinas: Unicamp, 1995.
• PÊCHEUX, Michel. Análise Automática do Discurso (AAD-69). In: GADET, Françoise; HAK, Tony (org.). Por uma análise automática do discurso Campinas: Ed. Unicamp, 1997.
• PÊCHEUX, Michel. O discurso: estrutura ou acontecimento. Trad. Eni Puccinelli Orlandi. São Paulo: Pontes, 2006.
• ROBIN, Régine. A memória saturada Trad. Cristiane Dias e Greciely Costa. São Paulo - Campinas: Editora UNICAMP, 2016.
• RODRIGUES, Daniella Lopes Dias Ignácio; SILVA, Jane Quintiliano Guimarães. O ensino da escrita de artigo acadêmico na web: suas práticas discursivas e jogos de verdade. In: Fabiana Komesu; Juliana Alves Assis. (Org.). O ensino da escrita de artigo acadêmico na web: suas práticas discursivas e jogos de verdade. Belo Horizonte: PUC Minas, 2019, v. 1, pp. 46-65. Disponível em: https://www.editora.pucminas.br/obra/praticas-discursivas-em-letramento-academico-questoes-em-estudo-volume-1 Acesso em: 23 de fev. 2025.
  » https://www.editora.pucminas.br/obra/praticas-discursivas-em-letramento-academico-questoes-em-estudo-volume-1
• SCALCO, Rebeca. Desvendando a inteligência artificial: o guia definitivo para alavancar suas pesquisas acadêmicas. [s.n.].
• STIEGLER, B. La Technique et le Temps - 1. La Faute d’Épiméthée. Paris: Fayard, 2018.

• Editors in Charge
  Luciana Salazar Salgado (Universidade Federal de São Carlos - UFSCar)
  Lucia Santaella (Pontifícia Universidade Católica de São Paulo - PUC-SP)
  Tony Berber Sardinha (Pontifícia Universidade Católica de São Paulo - PUC-SP)

Review

About the reviewerSCIMAGO INSTITUTIONS RANKINGS

Review

The article presents a current and relevant discussion for Discourse Analysis studies, providing pertinent bibliography and an appropriate theoretical approach, although it could include current references that have already addressed this topic, such as Gallo (2023) and Fernandes (2024), among others. Three changes are necessary for publication: 1. The title states that there is authorship in AI, which is questioned throughout the text until a conclusion is reached. Perhaps the issue here is authorship and not “danger,” an unusual term in scientific discourse that does not refer to a complement: fear of what? 2. Instead of the term “epistemic authorship,” I suggest changing it to “epistemic author position,” as this better fits the definitions provided by the author. 3. The expression “proletarianization of the consumer” requires further development. Where does the term come from? Why would becoming a proletarian be directly related to a “passive receiver”? In Fernandes (2024), https://revista.abralin.org/index.php/abralin/article/view/2183, there is mention of a “writing position” that refers to this apathy or lack of positioning of the subject. I suggest familiarizing yourself with this previous discussion before proposing new terms. APPROVED WITH SUGGESTIONS [Revised]

• peer review recommendation: accept

History

• Peer review received
  04 May 2025

Publication Dates

• Publication in this collection
  21 Aug 2026
• Date of issue
  2026

History

• Received
  23 Mar 2025
• Accepted
  07 Aug 2026