1)Have I used generative AI in a fashion to ensure that theprimary ideas, insights, interpretations, and critical anal-yses my own? (2) Have I used generative AI in a fashionto ensure that humans will maintain competency in coreresearch and writing skills? (3) Have I double checked toensure that all the content (and references) in my manu-script are accurate, reliable, and free of bias? and (4) HaveI disclosed exactly how generative AI tools were used inwriting the manuscript, and which parts of the manu-script involved the use of generative AI?

The importance of the above exercise is not that it pro-vides an exhaustive list of ethically permissible andimpermissible actions.

As noted above, published evidence indicates thatChatGPT is notoriously unreliable in citing referencesand can easily generate grammatically perfect refer-ences that do not actually exist.

. In each case, these uses taskChatGPT with generating novel text and thus haveheightened potential for the introduction of bias, hal-lucination, or plagiarism if engaged in uncritically.

However,some limitations are that ChatGPT is unable to accu-rately interpret the subtle nuances between languages,slang words, or cultural terms, which sometimes leads toinaccurate translation [2

The mostethically acceptable tier includes those items in whichChatGPT is primarily used to re-structure previouslyexisting text or ideas, including grammar and spelling,readability, and language translation

It is a concern, then, if scholars become too dependenton ChatGPT for ideation, generation of primary writtencontent, and initial data interpretation, as this assistancecould easily devolve into a dependency that arrests fur-ther scholarly development

There are no ‘fact-checking’ compo-nents to this process in current models, which is animportant source of the AI hallucinations and propa-gation of biases within the training dataset

We found that Chat-GPT oversimplified many items in this list, which mayresults in misleading guidance for academic authorsand scholars

Research-ers ultimately hold full responsibility for the originalityof their manuscript, accuracy and relevance of content,and appropriate referencing of published literature

Anotherstudy found that amongst 30 short medical papers gener-ated by ChatGPT, nearly half of the references were fab-ricated, while 46% of the references were authentic butinaccurate [16].

A final issue is that AI tools will tend to generate out-put that reproduces or amplifies biases inherent withinthe source data [19].

other articles (i.e. double plagiarism) [5].‘AI Hallucination’ is a phenomenon where the genera-tive AI tool offers convincing, but completely fabricated,nonsensical, or factually inaccurate content in responseto prompts provided by the user [13–15].

Left unchecked,this may result in a snowball effect, where authorsunknowingly cite AI-generated plagiarised text publishedin other articles (i.e. double plagiarism

Generative AI tools access publicly available data togenerate text, which may produce content that closely (orexactly) resembles the original source of information

Studies highlight three keychallenges with content created by LLM-based genera-tive AI tools: (a) plagiarism [5, 6]; (b) fabricated or falseinformation (i.e. ‘AI hallucination’) [13–15]; and (c) inac-curate or fabricated references [13–17]

While generative AI tools can help tosupport and improve the academic writing process, theymust not be used in place of the contributions of the aca-demic research team

Another core principle reflected in these statements isthat, despite their ability to generate novel and new text,generative AI tools do not meet the generally agreedstandards of authorship of academic research

We argue that the mosthelpful and transparent means for describing the useof these tools is within the methods section of a manu-script, where authors are encouraged to outline how AItools were utilised and how AI-generated output washandled (e.g. Was the AI-generated text reviewed or dis-cussed? Was the text edited? When relevant, were origi-nal sources of information identified?) and incorporatedinto the final manuscript.

Firstly, and perhaps mostimportantly, transparency about the use of generative AItools in the academic writing process is a cornerstoneof academic integrity [7, 9, 11].

Academic journals and their publishers have a vestedinterest in ensuring the integrity of the research thatthey publish and so explicitly publish their own policieson academic integrity and ethical practice in academicpublishing

To better understand the scope of recommendationsprovided by journals and publishers about the use ofgenerative AI in academic writing, we conducted a lit-erature search and also asked ChatGPT4 the question:“What recommendations exist for ethical use of Chat-GPT in academic writing?

We arguethat it is possible to use generative AI tools to supportthe academic writing process, and that doing so is alegitimate and natural part of the evolution of the globalhealthcare simulation research community, as long ascertain ethical safeguards are adhered to.

Unlike these previous tools, however, generativeAI tools have introduced a challenge to traditionalnotions of academic integrity: rather than simply help-ing researchers more efficiently conduct or completeresearch tasks, generative AI tools now have the ability toproduce novel written content on their own

ChatGPT represents a cat-egory of generative AI known as large language models(LLMs), which are indexed catalogues of text pulled fromhuman-generated content, designed to build coherentresponses to increasingly complex questions based on thestatistical relationships inherent in their training dataset

We first explore how academic publishers arepositioning the use of generative AI tools and then describe potential issues with using these tools in the academicwriting process.

Despitetheir tremendous potential, studies have uncovered that large language model (LLM)-based generative AI toolshave issues with plagiarism, AI hallucinations, and inaccurate or fabricated references

Generative artificial intelligence (AI) tools have been selectively adopted across the academic community to helpresearchers complete tasks in a more efficient manner.

However, these advantages are accompanied by challenges related to technical accuracy, ethicalauthorship, and standardized writing styles

Integrating AI in scientific writing offers significant benefits, including enhanced productivity,improved access for non-native English-speaking researchers, and optimized literature reviews.

Continuous supervision and rigorous ethical practicesare essential to ensure that AI solidifies as a robust ally in scientific writing, preserving the quality,originality, and integrity of academic contributions while preventing potential issues related toplagiarism (17).

While a powerful tool, AI should complement human work rather than replace it, there havealready been suggestions to use the term "co-intelligence" instead of "artificial intelligence"(15).

Authors are expected to disclose any use of AI tools during the preparation,writing, or revision of manuscripts, with details included in both the abstract and methodologysections.

Ethical issues play a central role in the use of AI in scientific writing(10).

The standardization of writing style observed in ChatGPT can restrict the creativity and individualexpression of authors(9).

For non-native English-speaking researchers, toolslike ChatGPT and Grammarly have become true partners, helping to overcome language barriers andimprove text clarity and coherence, making scientific publications more accessible and impactfu

ools such as Perplexity and Consensushave proven excellent allies in literature search and analysis, helping scientists quickly and accuratelyidentify trends and gaps in scientific production.

The final step involves evaluating whether the use of AI complies with ethicalguidelines

Decision tree

A decision tree was developed to further illustrate the decision-making process of AI-assistedscientific writing (

Grammarly efficiently identified grammatical errors, improved cohesion, andenhanced stylistic consistency.

ChatGPT and Grammarly demonstrated strong support in various aspects of scientificwriting. ChatGPT effectively generated drafts, synthesized findings, and suggested innovative topics,particularly for introductory and summary sections.

, its responses lacked depth in more complex queries, requiringcomplementary searches for comprehensive analysis.

Perplexity stood out for its ability to search for updated information and produce concisesummaries

Elicit effectively synthesizes information from multiple articles, generating precisesummaries without requiring full readings

This study revealed significant benefits of using artificial intelligence (AI) tools in scientificwriting.

Grammarly reviewed text, focusing on grammar correction, improvingtextual cohesion, and detecting plagiarism.

The evaluation of AI tools followed specific stages, using criteria such as synthesis capability,accessibility, and information accuracy.

hese issues underscorethe critical importance of human oversight and the need for well-defined ethical guidelines to ensurethat the use of AI in scientific writing remains rigorous, reliable, and aligned with academicstandards

This study employed a narrative analysis based on a non-systematic literature review,complemented by the author's experience in using artificial intelligence (AI) tools for scientificwriting.

research gaps

However, despite these significant advantages,notable obstacles remain, such as the risk of excessive standardization of language and the potentialfor technical inaccuracies that could affect the integrity of scientific writing.

AI-drivenplatforms like Perplexity, Consensus, and Elicit have become valuable resources

synthesis capabilities

The study concludes that while AI cansignificantly support scientific writing, its adoption should be accompanied bystringent human oversight and adherence to ethical guidelines to maintainacademic integrity.Key Words: Artificial intelligence,writing, dentistry, communicationbarriers, ethics.

In education, incorporating digital translanguaging strategies can support multilingual learners, fostering more inclusive teaching approaches

The findings have significant implications for sociolinguistics, offering insights into digital language practices and language evolution in globalized contexts.

As platforms continue to evolve and the global exchange of ideas grows, these practices will likely shape the future of digital communication and multilingual education.

The findings from this study confirm the adaptive and creative use of code-switching and translanguaging in digital spaces

The fluid use of multiple languages in digital environments challenges traditional monolingual pedagogies that often separate languages into distinct categories.

The prevalence of code-switching and translanguaging in digital spaces signals broader shifts in language evolution and offers valuable insights for multilingual education.

This was particularly evident on platforms like Instagram, WhatsApp, and YouTube, where users often integrate text with images, videos, and memes. For instance, captions in multiple languages, combined with visuals, allow users to share a richer narrative that appeals to their multilingual audience

On Twitter, where character limits and conciseness dominate, code-switching was more common (72% of posts) as users alternate between languages to convey their messages in a succinct manner.

Code-switching also serves as a means of efficiency, allowing speakers to convey personal sentiment concisely, especially on platforms like Twitter, where brevity is paramoun

Code-switching, which occurred in 68% of the analyzed posts, primarily emerges as a response to audience composition and platform norms.

The findings from this study offer a detailed exploration of how multilingual individuals engage in code-switching and translanguaging on social media platforms.

. WhatsApp and Instagram encouraged translanguaging due to their multimedia features and conversational nature. Conversely, Twitter exhibited higher rates of code-switching, attributed to its character limit and the need for concise communication

This flexibility allowed them to fill lexical gaps or simplify communication without losing meaning.

Translanguaging, in contrast, thrived in multimedia contexts, where users blended languages to enrich storytelling and emotional expression.

Code-switching emerged as a key strategy for accommodating diverse audiences and adhering to the unique norms of various social media platforms. Users frequently switched between languages to connect with multilingual followers and ensure their posts resonated widely.

Code-switching appeared in 68% of the analyzed posts, predominantly influenced by audience composition and platform norms. Translanguaging was identified in 42% of posts, with a notable prevalence in multimedia content, such as videos, memes, and image captions

Qualitative data from the interviews underwent thematic coding to uncover recurring patterns and themes.

The coding process involved categorizing the posts based on language pairs, context of usage, and the apparent intent behind the language choices

Participants were asked to complete detailed questionnaires focusing on three key areas: language use, online habits, and motivations for engaging in code-switching and translanguaging.

The participants' linguistic backgrounds were intentionally varied, covering a wide range of language pairings and combinations

The study engaged a total of 120 participants between the ages of 18 and 35, all of whom were fluent in at least two languages.

As digital communication continues to evolve, these linguistic practices will play an increasingly important role in shaping the future of language and communication in a globalized world. III. METHODS

However, the influence of dominant languages like English continues to shape online multilingual practices, creating tensions between linguistic diversity and the global hegemony of English in digital spaces

The digital space enhances this phenomenon, offering new opportunities for individuals to engage in multilingual practices that reflect both local and global affiliations.

Unlike code-switching, which often occurs at specific points of conversation or within certain boundaries, translanguaging enables speakers to draw upon their entire linguistic repertoire without the constraints of separating languages

For instance, bilingual individuals may switch from one language to another to signal a shift in topic or to evoke a certain emotion or cultural reference.

where elements like hashtags, character limits, and multimedia formats provide new ways for speakers to codeswitch and create hybrid linguistic forms (Androutso, 2015). These adaptations allow individuals to participate in and shape conversations in ways that would not have been possible in traditional face-to-face interactions.

Code-switching, the practice of alternating between two or more languages or dialects within a single conversation or utterance, has long been recognized as a crucial aspect of multilingual communication (Gumperz, 1982; Myers-Scotton, 1993)

the factors that influence the use of code-switching and translanguaging, such as the social context of communication, the relationship between interlocutors, and the medium of communication. It will also explore how these practices contribute to the construction of identity, the negotiation of meaning, and the maintenance of cultural ties in the digital age.

there remains a significant gap in our understanding of how these practices function and evolve in online spaces

Translanguaging, on the other hand, refers to the fluid and dynamic use of multiple linguistic resources to convey meaning, often in ways that transcend traditional boundaries between languages. It involves drawing on a speaker's full linguistic repertoire, including elements from different languages, dialects, and registers, to create meaning in context. Unlike code-switching, which typically involves the use of distinct languages or dialects, translanguaging emphasizes the seamless integration of linguistic resources to facilitate communication

allowing users to switch between languages with greater ease and frequency than ever before. In online environments, code-switching can occur in various forms, such as mixing languages within a sentence, switching languages between different parts of a conversation, or even alternating between different registers or varieties of a single language.

Code-switching refers to the practice of alternating between two or more languages or dialects within a single conversation, often depending on the social context, topic, or interlocutor.

The advent of online platforms, social media, and messaging apps has facilitated new avenues for interaction, allowing individuals from diverse linguistic backgrounds to engage with each other on a global scale. One of the most notable developments in this digital era is the increasing use of multilingual communication practices, which enable users to navigate between languages effortlessly.

using craft features effectively in a piece of writing tells the reader that you know your focus, and you are using craft as support for your larger idea–some people call it theme, some people call it a universal experience.

prose

Even in everyday writing activities, you identify your readers’ characteristics, interests, and expectations before making decisions about what you write.