Research Communication in the Age of AI

AI can now produce fluent academic text within seconds. So why should researchers still learn academic writing? This article argues that academic writing is not simply text production. It is the process of making research communicable — clear enough to be understood, trusted, evaluated and built upon. In the age of AI, that skill becomes more important, not less.

1. The question researchers are now asking

Recently, during a writing course, a participant asked me a question that I have been hearing more often: “Why are you still teaching academic writing? AI can write papers now.” It was not meant as a provocation. It was a genuine question, and I understand why researchers ask it.

Generative AI can now produce fluent academic text within seconds. It can draft abstracts, summarise literature, suggest titles, improve sentences, restructure paragraphs and generate something that looks, at least at first sight, like a journal paper. If academic writing were mainly about producing a text that resembles a paper, then the question would be difficult to dismiss.

But the question usually carries a second assumption as well. It is not only that AI can now write. It is also that journal papers themselves may be losing their central position. Researchers are increasingly visible on LinkedIn, podcasts, newsletters, YouTube, Instagram and TikTok. Some build public audiences that are much larger than the readership of many journal articles. Others use social media to explain their work, comment on developments in their field, or reach audiences who would never search for their publications in a database.

So the question is not only whether AI changes academic writing. The larger question is whether the ways researchers communicate their work are changing so fundamentally that traditional academic writing support no longer matters in the same way. My answer is that academic writing support matters very much — but only if we understand what academic writing is really for.

2. Why the debate about AI and journal papers can mislead us

The discussion about academic writing is often framed as a choice between old and new forms of communication. On one side, there is the traditional journal article. On the other, there are AI-generated texts, LinkedIn posts, podcasts, videos, newsletters, TikTok clips and whatever new formats will emerge next. It can easily sound as if the future of research communication will be decided by which medium replaces the journal paper.

A recent Nature editorial (2026) captured this wider shift very clearly. Under the title “The future of science communication is not an article like this”, Nature argued that science communication is becoming faster, more visual, more personal and more strongly shaped by social media platforms and artificial intelligence. The fact that Nature has joined TikTok is a strong signal that these developments can no longer be dismissed as peripheral.

I think this shift matters. Researchers, publishers and universities should not assume that knowledge will automatically travel through the channels of the past. Visibility, attention and trust are now shaped in new ways, and research communication has to respond to that.

But the debate can mislead us if we start with the medium rather than the purpose. We ask whether journal articles are dying, whether researchers should be on TikTok, whether LinkedIn posts are becoming more important, or whether AI will replace academic writing. These are relevant questions, but they are not the first questions.

Before we decide which medium researchers should use, we need to ask what research communication is supposed to achieve. A journal article, a conference talk, a policy brief, a LinkedIn post, a podcast or an AI-assisted summary are not ends in themselves. They are different ways of helping research reach an audience.

This is why the debate about AI and journal papers can become too narrow. If we only ask whether AI can write papers, we reduce academic writing to text production. If we only ask whether journal articles will be replaced by social media, we reduce research communication to platform choice. In both cases, we miss the deeper issue: research communication has a purpose, and the medium should serve that purpose rather than define it.

3. Academic writing is not text production

The reason academic writing still matters is that it has never been only about producing text. Of course, writing results in text. A paper, proposal, thesis chapter or conference abstract has to be written in words. But the visible text is not the whole process. It is the outcome of a deeper intellectual activity.

Academic writing is the process of making thinking communicable. It takes something that may still be complex, uncertain, incomplete or specialised, and shapes it so that another person can understand it, evaluate it, trust it, question it, use it and build on it. That process includes wording, structure and style, but it does not begin there. It begins with intellectual clarity.

This is why academic writing is often difficult. The difficulty is not only that researchers have to find elegant sentences. The difficulty is that they have to decide what their research actually means, what problem it addresses, what contribution it makes, what evidence supports it and why someone else should care. Writing forces these decisions to become explicit. It exposes gaps in the argument, unclear claims, weak transitions and conclusions that do not yet follow from the evidence.

AI changes the visibility of this problem because fluent text can now be produced very quickly. A draft may look polished before the thinking behind it is clear. A paragraph may sound confident even when the argument is weak. A paper may appear complete even when the contribution is still vague. This is why text production alone is not a reliable sign of good academic writing.

The central challenge is therefore not whether researchers can produce more text. Academia already produces a great deal of text: papers, reports, proposals, summaries, posts and outputs. The more important question is whether that communication matters. Does it clarify what was found? Does it explain why it matters? Does it help the right audience understand the work? Does it strengthen knowledge rather than merely add to the noise?

When academic writing is understood in this way, it becomes much more than a technical publishing skill. It becomes a way of clarifying research for oneself and for others. It helps knowledge move from the researcher’s mind to the audience’s mind in a form that can be examined, trusted and used. That is why writing remains valuable, even when tools for producing text become more powerful.

4. AI changes writing tactics, not communication principles

AI will certainly change academic writing. It already has. Researchers are using AI to move from rough notes to a first draft, revise unclear passages, translate, summarise, reorganise and test alternative formulations. For many, especially those writing in a second or third language, this can be genuinely helpful.

This is not something academia can simply avoid. AI will be part of how researchers search literature, read papers, prepare drafts, revise manuscripts and communicate findings. It will also influence how audiences discover and filter research. The question is no longer whether AI will enter research communication. The question is whether researchers can use it responsibly and judge what it produces.

The important distinction is that AI changes the tactics of writing, not the purpose of research communication. It can help produce a draft, but it cannot decide what the research means. It can improve a sentence, but it cannot know whether the argument deserves attention. It can suggest a structure, but it cannot take responsibility for whether that structure genuinely helps the intended audience understand the work.

The core communication questions remain with the researcher:

Who is this for?
What does the audience already know?
Why should they care?
What is the central contribution?
What evidence supports it?
What can be claimed, and what cannot?
What should the reader understand differently after engaging with this work?

These questions existed before AI, and they will still exist after the next generation of AI tools appears.

This is why AI should be seen as a powerful support tool, not as a substitute for research communication judgement. Used well, it can help researchers explore options and work more efficiently. Used poorly, it can produce fluent but shallow communication that sounds convincing while hiding weak logic, unclear contributions or unsupported claims.

For researchers, the deeper challenge is not simply to learn how to prompt AI. It is to know what good research communication requires, so they can guide, evaluate and revise AI-supported text with purpose.

5. Why knowledge and judgement still come first

This is where one of the most important points in the current AI debate becomes visible. In a recent LinkedIn article on generative AI and assessment, Wim Vanhaverbeke (2026) argued that if the artefact can be generated, we need to pay more attention to the person behind the artefact. If AI can produce the visible output, then we need to know whether the human being has the knowledge, judgement and understanding required to evaluate that output.

That argument applies directly to academic writing and research communication. If AI can generate a text, then the decisive question is no longer whether a text exists. The decisive question is whether the researcher can judge whether the text communicates the research well.

Researchers need knowledge of their subject, because only then can they recognise weak reasoning, hallucinations, false claims or misleading conclusions. But they also need knowledge of communication, because only then can they recognise whether a paper, proposal, presentation, post or AI-generated draft actually fulfils its purpose. Subject expertise helps researchers judge the scientific content. Communication expertise helps them judge whether the content has been communicated clearly, responsibly and meaningfully.

This distinction matters because fluent language can create a false sense of quality. A text may sound polished while the argument remains shallow. It may use the right academic phrases while the contribution is unclear. It may appear well structured while the logic does not quite hold. It may look like a finished paper while the researcher has not yet made the difficult decisions about audience, relevance, evidence and claim strength.

This is why AI does not reduce the need for expertise. It increases the value of judgement. Researchers who understand their subject and understand research communication can use AI as a useful partner in the process. They can ask better questions, give better instructions, recognise weak outputs and revise with purpose. Researchers without that judgement may simply accept fluent text because it sounds plausible.

For this reason, academic writing support cannot become only AI tool training. Researchers certainly need to learn how to use AI responsibly. But they also need to understand what good research communication looks like, so that they can decide whether AI-supported writing is doing its job. Knowledge comes first because judgement depends on it.

6. Five design principles for research communication

If the medium should follow the purpose, we need a clearer way to define that purpose. At TRESS ACADEMIC, we think successful research communication should make research visible, accessible, meaningful, credible and valid.

Visible means that the people who need the research can find it. A paper can be formally published and still remain practically invisible if it never reaches the relevant audience. This is one reason why additional formats such as conference talks, newsletters, social media posts, videos or policy briefs can be valuable.

Accessible means that the intended audience can reach and understand the communication without unnecessary barriers. This is partly about practical access, but it is also about intellectual access. The audience needs enough guidance to follow the argument without the research being oversimplified or distorted.

Meaningful means that the communication explains why the research matters. Many academic texts report findings accurately but leave readers uncertain about the significance of the work. Good research communication helps the audience understand what has been learned and what can now be thought, done or investigated differently.

Credible means that the audience has reasons to trust the source, the evidence and the reasoning. This becomes especially important in fast-moving communication environments where attention does not automatically equal reliability. Peer review, source transparency, careful attribution and visible limitations all help create credibility.

Valid means that the communication remains faithful to the research itself. It should not overstate the findings, hide uncertainty, remove necessary caveats or turn a limited result into a broad claim. Good research communication is not only clear and engaging; it is responsible.

These five principles change the discussion. Instead of asking which medium is best in general, we can ask which medium — or combination of media — best serves these purposes for a specific audience and a specific piece of research.

7. Why journal articles are not obsolete, but not enough

The five design principles also help us avoid a false choice between journal articles and newer communication formats. Journal articles are not obsolete. They still perform important functions in academic life: they create a scholarly record, connect new work to previous research, expose claims to peer review, and provide a format that other researchers can cite, evaluate, challenge and build upon.

At the same time, journal articles are not enough. They can be slow, difficult to access, hard to read and poor at reaching audiences beyond a narrow specialist group. Many published papers remain almost invisible to the people who could benefit from them. This is why other formats are gaining importance: not because they automatically replace journals, but because they can serve communication purposes that journal articles often struggle to fulfil.

Kaltenbrunner et al. (2026) show why change in scholarly communication is so difficult. The peer-reviewed journal article is not just a neutral container for research. It is tied to infrastructures, reward systems, prestige hierarchies, peer review, publishing business models and academic career expectations. That makes it difficult to replace, but also important to understand.

The better question is therefore not whether the journal article will survive. The better question is which functions it performs well, which functions it performs poorly, and which other media can complement it. A journal article may remain the formal scholarly record, while a conference talk, policy brief, LinkedIn article, podcast or video helps the same research become more visible, accessible or meaningful for different audiences.

The future of research communication is probably not one format replacing another. It is more likely to be a combination of formats, each serving a different purpose. The challenge for researchers is to understand the purpose first, and then choose the medium accordingly.

8. What this means for researcher development

If academic writing is understood as research communication, then researcher development also needs to shift. It is no longer enough to teach researchers only the tactics of one format: how to structure an introduction, write an abstract, select a journal or respond to reviewers. These skills still matter, but they are not the whole picture.

Researchers also need to learn how communication works across formats. They need to clarify their contribution, identify the right audience, build a convincing argument, explain relevance, use evidence responsibly and judge whether a piece of communication has achieved its purpose.

This becomes especially important in the age of AI. Researchers will increasingly work with AI-supported drafts, summaries, translations and revisions. But to use these tools well, they need more than technical prompting skills. They need the judgement to decide whether the output is clear, meaningful, credible and valid.

For universities, graduate schools and research institutions, this means that academic writing support should not become narrower. It should become broader and more strategic. The aim is not simply to help researchers produce more papers. The aim is to help them communicate research clearly, responsibly and effectively — whatever tools and media they use.

9. The skill worth teaching now

Perhaps AI is not making academic writing obsolete. Perhaps it is forcing us to rediscover what academic writing was always meant to do. It was never just about producing papers. It was about helping knowledge move from the researcher’s mind to the audience’s mind in a form that can be understood, trusted, evaluated and used.

That is why academic writing still matters. Not because every researcher needs to produce more papers. Not because the journal article is the only serious form of communication. And not because AI should be kept out of academic work.

Academic writing matters because it helps researchers make their thinking communicable. Once they understand that process, they can apply the same principles to a paper, a talk, a grant proposal, a policy brief, a LinkedIn article, a video or a future format we do not yet know.

The medium will continue to change. The tools will certainly change. But the need for clear, meaningful, credible and valid research communication will remain. That is the skill worth teaching now.

10. A practical next step for researchers and institutions

For individual researchers, one useful question is not only: “Is this text well written?” A better question is: “Does this communication achieve its purpose?” Before revising the next paper, proposal or presentation, it is worth checking whether the research is visible, accessible, meaningful, credible and valid for the audience it is meant to reach.

For universities, graduate schools and research institutions, the same shift matters at programme level. Academic writing support should not only help researchers produce manuscripts. It should help them understand how research communication works across different formats and how to use AI-supported tools responsibly without losing judgement, clarity or ownership of the argument.

At TRESS ACADEMIC, this is how we approach academic writing and publishing support. We help researchers prepare journal papers, but we also help them understand the deeper process behind successful research communication: how research becomes clear, meaningful, credible and useful to others.

#160: Research Communication in the Age of AI: Why Academic Writing Still Matters