Generative AI and the future of writing for publication: insights from applied linguistics journal editors

2.1 The role of journal editors

Journal editors and editorial staff, encompassing roles such as (co) editors-in-chief, managing editors, associate editors, section editors, and book review editors, are pivotal to the academic publishing process (Silver et al. 2023). The Committee for Publication Ethics (COPE), whose membership includes numerous journals, has outlined the essential responsibilities of editors in its Code of Conduct and Best Practice Guidelines for Journal Editors (COPE 2011). This document establishes that editors bear ultimate accountability for all content published in their journals. To fulfil this obligation, editors are required to:

1. Strive to meet the needs of readers and authors;

2. Continuously enhance their journals;

3. Implement processes to ensure the quality of published material;

4. Champion freedom of expression;

5. Maintain the integrity of the academic record;

6. Prevent business needs from compromising intellectual and ethical standards;

7. Publish corrections, clarifications, retractions, and apologies as necessary.

Beyond these mandatory standards, the guidelines delineate best practices, underscoring the multifaceted nature of editorial responsibilities (COPE 2011). Editors serve as gatekeepers, ensuring rigorous quality control by filtering out substandard submissions, selecting expert reviewers, and making publication decisions. They oversee the peer review process, liaising with reviewers and authors to enhance manuscript quality. Additionally, editors collaborate with publishers, societies, and editorial boards to set the journal’s strategic direction, including the formation of editorial boards, articulation of journal vision and mission, and development of policies governing manuscript preparation, declarations, and peer review. These responsibilities mean they cannot avoid the impact of GenAI on academic publishing (Casal and Kessler 2023). Publisher and journal policies about AI use, drafted after the realisation of the impact of ChatGPT on academic publishing, state their responsibility for safeguarding the curation process. For example, Elsevier’s AI policy for journals, while explicitly stating editors cannot use AI for any aspect of their editorial duties, also states, “responsible and accountable for the editorial process, the final decision and the communication thereof to the authors” (Elsevier, para.4, 2025). These responsibilities make them accountable for devising or implementing GenAI policies, but also, ensuring authors abide by the publisher and journal’s policies on use of GenAI. If questionable uses of AI are found in published papers, this could tarnish the journal’s reputation and possibly lead to retractions, leading the field to question the editor’s competence. The number of retractions of AI-generated articles is rising (Van Noorden 2023). Whether applied linguistics journal editors feel confident and competent with this responsibility is not understood.

The perspectives of applied linguistics journal editors on GenAI are not widely known. However, there was one survey study by Silver and colleagues (2023) that explored their views regarding research ethics and academic publishing. Their findings identified a shortage of expert reviewers as the most pressing ethical issue, with additional concerns including plagiarism, redundant publications, and inappropriate citations. Lesser issues included editorial interference and image manipulation. Interestingly, the survey revealed gaps in editors’ awareness of specific ethical challenges, such as unacknowledged writers and inappropriate acknowledgements, indicating the need for more comprehensive guidance. Notably, the study by Silver et al. (2023) did not address the role of artificial intelligence (AI) in academic publishing, as its emergence as a significant factor was just beginning. A recent editorial invited various technology education journal editors to share their voices on academic integrity in the GenAI era (Tlili et al. 2025). While highlighting the complex picture AI creates for editors, their main call was for more coherent and practical guidance for authors, reviewers, and editors on the use of GenAI, rather than, as they currently see it, ‘vague advice’. In addition, they called for journals to develop guidelines and regulations that reflect the nature of their research and writing conventions.

2.2 Artificial intelligence and academic publishing

Generative Artificial Intelligence (GenAI) represents a transformative subset of AI technologies explicitly designed for generative tasks (Chan and Colloton 2024). Large Language Models (LLMs) like ChatGPT, can perform a diverse range of natural language processing tasks without explicit pre-training, offering users unprecedented speed and contextual interactivity (Ali et al. 2024; Moorhouse 2024). This makes them capable of performing many tasks within the research and writing process that would have traditionally required humans. GenAI is increasingly embedded in widely used digital platforms such as writing processors (e.g., Google Docs), academic databases (e.g., SCOPUS AI), and data analysis software (e.g., MAXQDA) (Mizumoto and Teng 2025). Specialist tools like Consensus, an AI-driven academic search engine, illustrate the growing sophistication of these technologies in advancing research efficiency.

The capabilities and rapid adoption of GenAI have disrupted established norms in academic publishing, particularly concerning integrity and trustworthiness (Tlili et al. 2025). The primary focus has been on potential risks, including misconduct and the erosion of academic integrity in the research and publication processes. For instance, Casal and Kessler (2023) found that applied linguistics scholars and journal editors struggled to differentiate between ChatGPT-generated abstracts and human-authored ones, underscoring the challenges of detecting AI-generated content. However, there have also been concerns about bias and the inability of LLMs to discern nuances within academic texts – potentially reducing diversity within writing for publication (Kuteeva and Andersson 2024) and lead to ‘cultural nudging’ where some world views are more evident in AI responses than others (Hetzscholdt 2024). Scholars, such as, Stahl and Eke (2024) believe the ethical risks of GenAI outweigh the benefits it affords. Issues such as ‘hallucinations’ where GenAI tools create factually inaccurate information could undermine our knowledge base. Indeed, GenAI tools do not understand the data they are accessing or creating, and can present misleading, non-sensical or bias information (Dumit and Roepstorff 2025). In fact, their reliance on statistical probability means they can amplify bias, or misinformation, if they are dominating in their datasets (Kuteeva and Andersson 2024). While efforts to detect GenAI content have been made, detection tools have proven unreliable. Perkins et al. (2024) demonstrated how simple strategies could bypass AI detectors, creating the potential for false accusations and undetected breaches of publication ethics.

In response to the developments of GenAI, COPE issued a position statement on GenAI tools, stipulating that AI cannot meet the criteria for authorship because it cannot take responsibility for scholarly work (COPE 2023). The guidelines require authors to transparently disclose their use of AI tools in the Materials and Methods or equivalent sections of their manuscripts. They also hold authors accountable for the integrity of content generated with AI tools, emphasising the need for responsible use. Yet, the guidance from journals has been inconsistent (Tlili et al. 2025). Bagenal (2024) highlights this inconsistency, noting that only 24 % of the top 100 global academic publishers provide explicit guidance on GenAI use, and fewer than half of these require authors to disclose their use of AI. Similarly, Ganjavi et al. (2024) report that while 87 % of leading scientific journals offer guidance, only 43 % outline specific disclosure criteria for GenAI use.

Even where policies exist, ambiguity persists. Authors often fear that disclosing their use of GenAI may negatively impact their submissions. Farangi and Nejadghanbar (2024) identified this concern in their study of QRPs among Iranian applied linguists. Participants expressed reluctance to disclose their use of AI tools, citing fears of biased peer review and rejection. One participant noted that declaring AI-assisted data analysis led to prolonged peer review, highlighting the potential for editorial bias against GenAI usage. Tan and colleagues (2025) found that people do judge texts they considered to be generated by GenAI more critically than texts they assume to be written by humans. The lack of clarity exacerbates ethical concerns, particularly around unreported GenAI use, which risks undermining trust in academic publishing (Belcher 2025). At the same time, the integration of GenAI into editorial work, such as screening, peer review, or adding editorial decisions, could harm academic relationships, as editorial work is essential in shaping academic communities and norms (Hoessini and Horbach 2023).

The picture is not all negative. As has already been discussed, there are real and tangible benefits GenAI can bring to the research process. Studies by Morgan (2023) and Naeem et al. (2025) show that with human oversight, GenAI can assist with qualitative data analysis. They could increase rigor by being tasked with secondary data coding along with trained human coders. They may be able to analyse large amounts of data that humans may struggle to process (Mizumoto and Eguchi 2023). For example, they could be trained to review large numbers of abstracts or other data for keywords or patterns. They can accurately and contextually translate texts (Guo et al. 2024). Indeed, a few authors have begun to document their process of integrating GenAI into their research processes (see Moorhouse and Kohnke 2024). Regarding writing, they can help authors present their work more clearly and coherently, providing greater access to journals by less proficient users of English. They can also increase access to academic content through multimodal or multilingual representations (e.g., podcasts, infographics, translations) (Yeo et al. 2025). Ebadi et al. (2025) explored the perspective of twelve journal reviewers and found tangible benefits of using GenAI in their peer review processes. These included automating tasks, such as, preliminary screening, plagiarism detection, and language verification. These uses could reduce workload and enhance consistency in applying review standards. By acknowledging and integrating these advantages into editorial practices, journals could develop policies that balance mitigating risks with leveraging AI’s capabilities (Tlili et al. 2025).

2.3 Addressing the research gap

To navigate these challenges and capitalise on the advantages of GenAI, applied linguistics requires discipline-specific guidance on the ethical and practical use of GenAI in academic publishing (Farangi et al. 2024). This study addresses a critical gap in the literature by examining how editors of applied linguistics journals perceive and respond to these developments. By exploring their perspectives, this research aims to contribute to the development of transparent, equitable, and actionable policies that uphold academic integrity while embracing the opportunities presented by GenAI.

3.1 Participants

The study involved ten editors from applied linguistics journals. Participants were selected using purposeful sampling to target individuals with the most relevant expertise. The inclusion criteria were:

1. Holding a senior editorial position (e.g., editor-in-chief, associate editor) in a peer-reviewed applied linguistics journal.

2. Having a minimum of two years’ experience in editorial roles to ensure familiarity with editorial processes and exposure to submissions involving GenAI tools.

Recruitment was conducted via email, with invitations outlining the study’s purpose and procedures. The sampling strategy involved drawing on the three authors’ familiarity and connections with applied linguistics editors of various journals (Cohen et al. 2018). We started with our inclusion criteria and then brainstormed editors we knew professionally through our roles as editors, reviewers, and authors. We developed a list of 15 editors who represented a diversity of geographical locations, journal types, and prestige (indicated by Social Science Citation Index [SSCI] and SCOPUS indexing). We contacted all 15 editors. Of the fifteen, ten consented to participate. They represented various journal types, including academic societies, commercial publishers, and university presses (see Table 1 for demographic details). The journals they edited were based in North America, Europe, Australasia, and Asia. All but one were indexed in SCOPUS, with eight indexed in SSCI. Specific details of each editor’s journal are not provided to protect their anonymity (Cohen et al. 2018). As we progressed through the data collection process, we observed a similar pattern in the participants’ responses to our interview questions. Therefore, we believed the sample size was sufficient to achieve data saturation, a standard benchmark in qualitative research for identifying recurring themes (Cohen et al. 2018). However, in line with calls to make researcher subjectivity and life capital visible within applied linguistics research (Consoli 2022), we acknowledge that our own editorial and professional networks shaped participant recruitment and the types of perspectives captured in this study.

3.2 Data collection

Data were collected through semi-structured interviews conducted via Zoom or Microsoft Teams between August and September 2024. Semi-structured interviews offered the flexibility to explore participants’ perspectives while ensuring that all key themes were addressed consistently across interviews. While other methods, such as reviewing journal guidelines, would have provided valuable insights (see Yin et al. 2025 for an example of a cross-sectional study of GenAI guidelines in medical journals, as well as, Yin and Chapelle 2025 for an exploration of guidelines of applied lingustics journals), we decided to deploy interviews. Interviews could provide deep insight into the editors’ thinking as well as their direct experience in relation to GenAI’s impact.

The interview guide was developed based on the research question and relevant literature with a focus on the following themes:

1. Awareness, understanding, and experiences of GenAI tools in academic writing and their editorial roles.

2. Perceived benefits and challenges of GenAI use in research and publication.

3. Appropriate and inappropriate uses of GenAI in research and publication.

4. Journal editors’ roles in relation to GenAI in academic publishing.

5. Policies or guidelines regarding GenAI within their journals.

To refine the guide, the lead authors reviewed the relevant literature, consulted the two collaborating authors, and uploaded a draft to GPT-4.0 for feedback. The prompt used was: “I am conducting a research project exploring applied linguistic journal editors’ perspectives of the use of GenAI in academic publishing. Can you provide some feedback on the clarity of the questions and the comprehensiveness of the guide?” Suggestions regarding overlapping questions, diverse GenAI experiences, additional questions, and clarifying concepts were reviewed and incorporated. For example, it suggested several of the questions were overlapping. It stated, “There is some overlap in questions about perceptions (e.g., Q3, Q4, Q10, Q11) and challenges (e.g., Q6, Q13). Streamlining these could make the guide more concise.” These were reviewed, and appropriate changes were made (see Table 2). Human judgement, accountability, and oversight were prioritised (Tlili et al. 2025). It also suggested adding a field-specific question, “Do you think the use of GenAI in applied linguistics journals differs from its use in other academic disciplines? If so, how?”. The question was added to the final guide. The finalised interview guide can be found in Appendix 1.

Interviews lasted between 40 and 60 min and were audio-recorded with participants’ consent. Given the semi-structured nature, each interview followed a different flow. However, we covered the different aspects of our guide during each interview. Conducting interviews via video-conferencing software facilitated participation from geographically diverse editors. It accommodated their schedules (Guo et al. 2024). Audio recordings were transcribed verbatim using machine transcription software, with transcripts cross-checked by the researchers against recordings to ensure fidelity. To maintain confidentiality, identifiable information was removed, and pseudonyms were assigned to participants.

3.3 Data analysis

The study employed thematic analysis as the methodological framework for analysing data (Cohen et al. 2018). This approach is well-suited for identifying, interpreting, and reporting patterns within qualitative data and is widely utilised in exploratory research. To develop a coding scheme, the three researchers selected one interviewee’s transcript to code (Maggie). Each researcher read and re-read Maggie’s interview transcript and coded it independently. We used a combination of inductive coding – allowing themes to emerge organically from the data – and deductive coding guided by the study’s research question. We then met to discuss our coding. We combined our emerging themes into a table and compared them with each other. Discrepancies were discussed, and a final coding scheme was developed (see Table 3 for the initial coding scheme). However, we continued to use inductive coding on the remaining transcripts to ensure no relevant themes were ignored and left out.

After the meeting, we assigned two researchers to each remaining transcript. This allowed us to compare our coding. We continued to adapt our coding scheme as we analysed the data and re-visited previous coded transcriptions to identify salient patterns across participants. We then meet to identify our final themes (see Table 4 for the final themes with example codes and data extracts). The change in Table 3 to 4 shows how we engaged in a reflexive and iterative process throughout the data analysis stage (Braun and Clarke 2006). By incorporating reflexivity throughout the design and implementation of this study, from participant selection to theme development, we align with a growing recognition in applied linguistics that reflexivity is a hallmark of ethical and rigorous research practice (Consoli and Ganassin 2023).

4.1 The context of academic publishing

The editors consistently emphasised the need to situate GenAI within the broader challenges of academic publishing. The data indicate that systemic pressures – such as increasing institutional demands for publication output – are driving a surge in submissions, often at the expense of quality. Gary lamented “The market has driven people crazy to do everything they can to get published instead of really producing good knowledge.” He highlighted that many of the submissions he reviewed “did not add anything new to the field.” Similarly, Osman observed “an increase in submissions but not an increase in quality,” reflecting broader concerns about the diminishing rigour in academic contributions.

The strain on journal operations was also a recurring theme. Editors described difficulties in managing rising submission volumes and finding appropriate reviewers. Maggie noted, “We have an explosion in the number of articles… and then we don’t have enough reviewers.” Gary attributed this pressure in part to an “explosion of open access journals,” which require additional peer review resources. Ruth critiqued a publication-driven academic culture, stating, “Scholarship is about churning out articles [which] has affected generating knowledge.”

Despite these challenges, some participants observed emerging efforts to shift the paradigm. Osman pointed to examples of some European universities that have “abolished the rankings,” signalling a gradual move away from metrics-driven publication pressures. Within this complex environment, the editors saw their roles as custodians of quality, tasked with maintaining rigour, managing the peer review process, and steering their journals’ strategic direction.

4.2 The role of GenAI in exacerbating existing pressures

Editors expressed concern about how GenAI could intensify existing challenges in academic publishing. The ease with which AI tools can generate content risks exacerbating the tension between quantity and quality. Patrick voiced apprehension, noting that “Other editors are concerned because they have seen an increase in submissions, and they know that some of that might be due to the assistance [or] over-reliance on AI tools.” As Gary noted, this influx of submissions risks “contaminating the knowledge pool,” making it harder to discern meaningful contributions.

Moreover, the speed at which GenAI can generate manuscripts raised fears of overwhelming the peer review process. Ruth questioned whether editors could manage the volume: “People are worried about the pace… if someone is just churning out a bunch of articles with AI, then can we possibly keep up with checking all of those?”

These findings underscore a pressing need to address the implications of GenAI in applied linguistics publishing. The data suggest that while the technology holds promise for enhancing productivity, it also risks perpetuating systemic issues and compromising academic integrity. Editors are caught in the tension between enabling innovation and safeguarding the quality and rigour that underpin scholarly knowledge production.

4.3 The responses to GenAI in academic publishing

All the editors recognised the necessity of responding to the rapid development of GenAI in applied linguistics writing for publication. While they acknowledged the increasing sophistication of these technologies and their potential generative capabilities, most editors had not yet integrated GenAI into their editorial workflows. However, many had explored its utility in other professional domains. For example, Nova described using ChatGPT to develop rubrics based on assignment guidelines, Anson highlighted its use for supporting text translations in collaborative research, and Maggie reported employing GenAI tools “on a daily basis” in her teaching practices. In his article review work for another journal, Sat Wing mentioned using an AI tool to refine reviewer feedback, which he found “looked more professional.”

4.4 Concerns about GenAI in academic publishing

The editors raised several concerns about the integration of GenAI into applied linguistics research and publishing. These included fears about a growing prevalence of generic content, reduced originality, disengagement from the research process, challenges to research integrity, and issues of equity. Additionally, there was caution about how GenAI might be used in editorial work.

4.5 The potential and acceptable utility of GenAI in academic publishing

The editors’ responses highlighted considerable ambiguity regarding the potential and acceptable uses of GenAI in academic publishing. This uncertainty spanned various stages of the publishing process, including manuscript preparation, editorial work, and peer review.

5.1 Challenges of GenAI in an already-pressured landscape

The study shows that journal editors see the development of GenAI as exacerbating an already challenging publishing context. Applied linguistics journal editors are under a lot of pressure due to increased submissions and the challenge of finding qualified reviewers. This has put a strain on the manuscript curation process, but, as the editors mention, it also risks diluting the knowledge pool as more and more articles are added to the field. Journal editors are concerned that within the “publish and perish” culture (Kendall and Teixeira da Silva 2024; Yeo et al. 2025), applied linguistics scholars are incentivized to speed up their research activities and manuscript writing using GenAI tools. Yet, the study suggests that editors’ concern is not whether GenAI tools are used or not. They are concerned about the way GenAI tools are used, have ethical concerns regarding the use of GenAI tools, and have a preference for the transparent use of GenAI.

At the moment, the findings suggest that the publisher, journal policies, and the wider publishing sector are creating ambiguity towards the use of GenAI, which leads authors to take a cautious or deceptive approach and not declare their use of tools. Similar issues have been found in the higher education context (e.g., Gonsalves 2024; Tan et al. 2025). Indeed, although the editors had identified the use of GenAI in submitted manuscripts, they had not observed scholars declare their use of GenAI tools in their submissions. The reluctance to declare use is somewhat understandable if the submitting author is concerned their manuscript may not be treated fairly by the editor if they declare honestly about their use (Farangi and Nejadghanbar 2024). However, it speaks to wider issues of questionable research practices that are becoming prevalent in academic discourses/debates about publishing (Nejadghanbar et al. 2023). The editors highlighted that applied linguistics may be particularly at risk of undeclared GenAI due to the kinds of problems the field addresses and the methodologies used. If anything, it increases the scepticism and mistrust of scholarship in the field, damaging its reputation (Moorhouse and Nejadghanbar 2025; Szudarski 2025). As such, we argue that clarity is essential to ensure transparent use.

5.2 Ethical concerns and questionable research practices

The journal editors were concerned about the ways researchers may use GenAI. Using tools to support language accuracy seems to be uncontroversial. Despite scholars raising issues of language standardization and limited diversity in LLM responses (Hetzscholdt 2024; Kuteeva and Andersson 2024). However, other ways GenAI can be used, such as for synthesizing literature for the literature review or analyzing data (Belcher 2025), were met with more caution. Editors cautioned that these uses may reduce the researchers’ engagement with the process, potentially increase participant risks, and system biases could affect the studies’ findings. The understanding that AI use requires human oversight was paramount to the editors (Chen et al. 2024; Cotton et al. 2023; Dumit and Roepstorff 2025).

The cases described by Anson and Maggie, where authors submitted manuscripts with fictitious references and inaccurate information, highlight the potential for questionable research practices to become more pervasive with GenAI. These examples underscore the urgency of delineating acceptable and unacceptable uses of GenAI in research and publishing. Editors called for strengthening ethical training for researchers, particularly focusing on GenAI’s appropriate applications. This training should emphasise how responsible use of GenAI can improve research quality while discouraging practices that compromise the integrity of scholarly work (Ebadi et al. 2025; Farangi et al. 2024).

5.3 A call for discipline-specific guidance

The study highlights the need for applied linguistics editors to collaborate on developing discipline-specific guidance or a code of conduct that reflects the field’s diverse research traditions. Such efforts could include colloquia, panel discussions, and opinion pieces that lead to formal position statements on GenAI use. The discussion of GenAI use in academic publishing needs to draw on diverse stakeholders and consider the diverse methodological traditions of the field (Tlili et al. 2025). Applied linguistics research has unique characteristics that should be reflected in any guidance. As Chapelle et al. (2024) suggest, applied linguistics focuses “on language in context from relevant technical perspectives is a distinct value that applied linguistics research provides beyond methodologies offered by other fields of study” (p. 1). Therefore, guidance needs to recognise the problems the field addresses, the human-centred orientation and methodological focuses of the field and see GenAI as tools that enhance the study quality in the field, not just speed up or automate processes at the expense of research traditions and quality (Moorhouse and Nejadghanbar 2025).

In addition, there is a need for methodological models and protocols that detail how GenAI can be integrated into research and writing processes. This would provide authors with a way to document their process and give editors, reviewers, and readers the ability to assess the appropriateness of using GenAI tools in the submitted manuscript. Such models and protocols could lead to more fine-tuned guidance on using GenAI in applied linguistics research, which, in turn, could incentivize researchers to declare their use. In essence, moving from a deficit view of the use of AI to a more strategic view of the use of AI – as a way to enhance the quality and rigor of the research and writing processes (Morgan 2023; Naeem et al. 2025). While acknowledging the complexity of GenAI and writing and promoting ways AI could help open up research to broader audiences by demystifying publication practices (Kuteeva and Andresson 2024). By leading these developments, the field can establish principled uses of GenAI that align with the field’s rich and diverse research traditions, rather than relying on publishers or external stakeholders to dictate the terms.

GenAI represents a transformative force that demands action. Applied linguistics must balance its potential benefits – such as enhanced efficiency and improved language quality – with the risks it poses to research integrity and originality. Editors and researchers alike must collaborate to establish norms and frameworks that ensure GenAI’s responsible use. By proactively addressing these challenges, the field can leverage GenAI’s potential to advance knowledge while safeguarding the rigour and ethics of academic publishing.