Your Science, Your Voice

How IEL Supports Researchers in the AI Era

Artificial intelligence has moved from a novelty to a day-to-day writing tool remarkably quickly. Many researchers now draft sections of a manuscript with the help of AI, tidy their English using automated editors, or explore tools that summarise literature and speed up early brainstorming. In laboratories, offices, and home workspaces around the world, AI has simply become part of the writing ecosystem.

At Insight Editing London, our perspective is straightforward: researchers will use AI, and should feel able to do so safely, transparently, and without losing their scientific voice. The question is no longer whether AI will shape academic writing, but how researchers and editors can use it responsibly while preserving clarity, accuracy, and integrity.

AI is here as a tool, not an author

Major publishers are increasingly integrating custom AI-screening tools into their manuscript triage systems. These tools are not designed to “catch” AI use outright, but to flag text that may require human review — such as overly generic phrasing, formulaic structures, or inconsistencies that sometimes arise in AI-assisted writing. In many ways, these automated screens now function like an additional quality and integrity checkpoint.

This is good news. It means researchers can use AI tools where helpful, as long as the manuscript remains clearly authored, clearly human, and clearly grounded in the authors’ understanding of their own work. What matters is that:

• the scientific content remains the authors’ intellectual contribution
• the writing reflects the authors’ voice and intention
• the manuscript is accurate, transparent, and original
• AI has supported the writing process, not generated the reasoning behind it

AI assists — but it does not think, interpret, double-check, take ownership of ideas, or say “I don’t know”. And this is precisely where trusted editorial support becomes essential: to help ensure that AI-assisted text is accurate, authentic, and aligned with the standards of international scientific publishing.

Embracing AI without losing your voice

One of the most common concerns we hear from researchers — particularly those whose first language is not English — is that AI can flatten the tone of a manuscript. Text may become linguistically correct but stylistically generic; phrasing becomes repetitive; the narrative loses shape and clarity.

Researchers often tell us:

• “It sounds like everyone else.”
• “It no longer reads like my work.”
• “My supervisor said it feels robotic.”

This isn’t a failure on the researcher’s part — it is simply how large language models operate. They predict patterns, smooth variation, and remove the idiosyncrasies that make scientific writing human.

Our role at IEL is to help you to restore that individuality: your structure, your reasoning, your emphasis, and your way of communicating your science. Our editors help ensure AI-assisted text remains genuine, coherent, and aligned with the conventions of scientific writing — and with your authorial voice.

Embracing AI without losing your skills

A second concern raised by supervisors and journal editors is that an over-reliance on AI can slowly weaken the communication skills researchers need throughout their careers. Clear scientific writing strengthens critical thinking, deepens understanding of the work, and helps researchers communicate confidently with funders, collaborators, and the wider community. When AI does too much of this work, those essential skills — structuring arguments, explaining results, polishing narratives — are practised less often, and eventually weakened.

None of this means AI should be avoided. It simply means that researchers benefit from using it thoughtfully, as a tool that supports the writing process rather than replaces it. IEL’s role is to help maintain that balance.

How can IEL help? We believe AI should support researchers, not replace the skills that underpin strong scientific communication. Online training has always been central to our work and is one of the key ways we differ from other editing companies.

Through clear in-text comments and concise editorial reports, we explain what we adjusted, why it matters, and how the writing could be strengthened further. This approach not only improves the current manuscript but also helps researchers build confidence for future papers. In an era where AI can tidy language but cannot teach judgement, narrative flow, or discipline-specific conventions, our focus remains on helping authors strengthen — not sidestep — their communication skills.

Supporting researchers who feel uncertain about AI

In many regions, the pressures surrounding AI use are acute. Researchers aiming for international English-language journals often worry about:

• their text being flagged by automated AI-detection systems
• accidentally using AI-generated text too heavily
• producing language that is correct but feels non-native
• being accused of misconduct despite good intentions
• losing skills in scientific writing

These concerns are legitimate, and the more AI develops, the more researchers need clear, human-led guidance to navigate it safely. For that reason, IEL helps authors:

• check that AI-generated phrasing is accurate
• correct over-generalised or vague sentences
• balance fluency with authenticity
• develop writing skills rather than outsource them

Where human editorial expertise still matters most

While AI can tidy sentences, it cannot replace the analytical, interpretive, and ethical judgement that sits at the heart of scientific communication. These are areas where human editorial expertise remains essential:

• Scientific reasoning — interpreting results, explaining logic, making nuanced comparisons
• Accuracy and citations — AI may invent or misattribute references
• Overclaiming and hype — AI tends to overstate novelty or impact
• Consistency — in voice and use of discipline-specific conventions
• Ethical and integrity checks — authorship, transparency, image guidance, responsible reporting

We embrace this new era with optimism and realism. We are here to help you use AI confidently, preserve your scientific voice, and communicate your work with clarity, accuracy, and integrity. Your science deserves to be heard — in your words, in your way, and with the trusted support of experienced human editors.

At Insight Editing London, we see first-hand the challenges that researchers face in writing and publishing responsibly. Every week we edit manuscripts where authors are doing excellent science, but every now and then we come across examples where the way in which the science is written risks unintentionally undermining its integrity. And of course, there are also the wider systemic pressures: competition for funding, the pressure to publish quickly, and the rise of new challenges like AI and paper mills.

This month, we are highlighting some of the issues we encounter most often in scientific communication today. These aren’t abstract problems—you’ll recognise them in discussions in your lab, in peer review reports, and perhaps even in your own drafts. Our aim is to explain why these practices matter, what can go wrong if they are ignored, and how you can avoid the pitfalls.

Our aim here is not to cover every detail, but to raise awareness of the key issues that can affect integrity in scientific communication. To help you take this further, we’ve included links to trusted resources and guidelines where you can explore each topic in more depth.

1. Authorship & Contributorship: Giving Credit Where It’s Due

Questions about authorship are one of the most common integrity issues we hear about from clients. Should a lab head be included? What about the student who generated the data but has since left?

Gift authorship (adding someone with little or no input) and ghost authorship (leaving out someone who made a real contribution) both distort the record. They might feel like small compromises in the moment, but they create tension and unfairness.

We’ve seen the consequences: disputes between collaborators, strained relationships, and in some cases, corrections or even retractions. The best solution is transparency. Talk about authorship early, update agreements as the project develops, and use contributorship statements to spell out everyone’s role.

• ICMJE authorship criteria: https://www.icmje.org/recommendations/
• COPE resources on authorship: https://publicationethics.org/authorship

2. Attribution & Plagiarism: Respecting the Work of Others

As editors, we sometimes come across text that looks “borrowed.” Often it isn’t intentional plagiarism, but rather careless paraphrasing or recycling wording from an earlier paper. Still, it matters.

Plagiarism—whether direct copying, close paraphrasing, or recycling your own text without acknowledgement—misrepresents originality. Journals use detection software and take it seriously. Consequences can include rejection, retraction, or worse.

The solution is simple: cite properly, paraphrase carefully, and if in doubt, over-acknowledge rather than under.

• COPE discussion documents on plagiarism and text recycling: https://publicationethics.org/resources/discussion-documents/plagiarism-text-recycling
• Elsevier publishing ethics policies: https://www.elsevier.com/about/policies/publishing-ethics

3. Ethical Writing & Data Reporting: Telling the Whole Story

One of the issues we notice most in manuscripts is not outright misconduct, but overstatement. For example, presenting results as more definitive than they are, or ignoring contradictory findings. Sometimes this comes from enthusiasm—it’s natural to want to highlight your strongest results—but it can tip into misrepresentation.

Reviewers are quick to spot cherry-picking or claims that aren’t backed by the data. At best, you’ll be asked for revisions; at worst, trust in your work is damaged.

Our advice: follow reporting guidelines, present limitations openly, and write conclusions that reflect the evidence you have—not the evidence you wish you had.

• CONSORT (clinical trials): http://www.consort-statement.org/
• PRISMA (systematic reviews): http://www.prisma-statement.org/
• EQUATOR Network (reporting guidelines across disciplines):
• https://www.equator-network.org/

4. Paper Mills & Predatory Publishers: Protecting the Literature

The rise of paper mills and predatory publishers is something we all need to be vigilant about. We hear from researchers who have been approached with offers of guaranteed publication or authorship slots. It can be tempting under pressure, but it comes with serious risks.

These outlets undermine trust, waste resources, and can permanently damage your reputation. We recommend always checking journals through resources like Think. Check. Submit. or by looking at the editorial board and peer review process.

• Think. Check. Submit.: https://thinkchecksubmit.org/
• COPE statement on paper mills: https://publicationethics.org/papermills

5. Conflicts of Interest: Transparency Builds Trust

Conflicts of interest aren’t always financial. They can be professional, institutional, or even personal. But whatever the form, lack of disclosure erodes trust.

We sometimes notice missing or incomplete conflict declarations during editing, and we always flag it. It’s better to over-disclose than under-disclose—transparency reassures readers and reviewers.

• ICMJE conflict of interest disclosure form: https://www.icmje.org/disclosure-of-interest/
• COPE resources on conflicts of interest: https://publicationethics.org/conflicts-of-interest

6. AI Misuse: A New Frontier in Integrity

AI is the newest issue in integrity, and one we’re already seeing in manuscripts. We notice when text feels “AI-smooth”—grammatically perfect but oddly flat, with structural or logical weaknesses.

AI can be useful for refining sentences or cutting word counts, but it cannot interpret data or argue a case. Journals are clear that AI cannot be an author, and use must be declared. Our position is simple: use AI for small tasks if you wish, but never outsource the intellectual core of your paper.

• EASE recommendations on AI in scholarly communication: https://ease.org.uk/2024/09/recommendations-on-the-use-of-ai-in-scholarly-communication/
• COPE position on authorship and AI tools: https://publicationethics.org/guidance/cope-position/authorship-and-ai-tools

7. Peer Review Ethics: Safeguarding the Process

We also hear stories from clients about unfair or biased peer reviews, or about confidential data leaking from the process. Peer review relies on trust and professionalism.

If you’re invited to review, only accept if you have the right expertise, be transparent about conflicts, and above all, respect confidentiality.

• COPE Ethical Guidelines for Peer Reviewers: https://publicationethics.org/peerreview
• Wiley peer review resources: https://authorservices.wiley.com/Reviewers/journal-reviewers/index.html

Integrity isn’t just about avoiding misconduct—it’s about building credibility and impact. As editors, we can tell you that reviewers do notice when manuscripts are overstated, under-acknowledged, or poorly structured. And they also notice when they are clear, transparent, and responsible.

At IEL, we help researchers strengthen their manuscripts not only for clarity and flow, but also for integrity. Through our reports, comments, and ongoing dialogue, we aim to make every paper not just publishable, but also a genuine contribution to the field.

If you’re preparing a manuscript and want to ensure it’s not only well written but also ethically sound, get in touch—we’ll be pleased to help.

Remember when teachers told us to learn mental arithmetic because “you won’t have a calculator in your pocket everywhere you go”? They were wrong about the calculator—we all carry one now in the form of some sort of smartphone or tablet. But they were right about the principle: some skills are too fundamental to outsource.

Today, large language models are the calculators of the research world. They can generate text, refine sentences, and come up with new hypotheses to test. But they can’t replace the core skills every scientist needs, including how to write a paper and communicate your findings. Let’s find out why!

1. You’re the World’s Closest Expert

Writing a scientific paper isn’t just about communicating results—it’s part of the scientific process itself. To write well, you need to:

• Understand exactly what you did, why you did it, and how your methods led to your findings.
• Place your work in the context of the literature, making clear where it fits and what gap it addresses.
• Interpret your data with both precision and curiosity, exploring what it might mean for the next steps in your field.

Without this process, your science risks having less impact—not because the results are weaker, but because they aren’t fully understood, communicated, or connected.

2. The Leadership Link

If you’re aiming to lead your next research project, writing is not optional. It’s where you show that you can think as well as experiment; that you can shape the questions, not just execute the methods. A PI who can’t clearly interpret and argue for their work is a PI who risks being led by others, whether by co-authors, editors, or increasingly, by algorithms.

Remember: before your work is published, no one knows your data like you do. You’ve lived it—at the bench, at the bedside, or in the field. You’ve seen patterns emerge, wrestled with unexpected results, and debated interpretations with your colleagues.

That unique, in-the-moment insight is yours alone until the paper appears in print. If you hand over the task of writing to AI, you risk losing that expert voice and the subtle understanding that makes your science truly yours.

3. What AI Can—and Can’t—Do

Now, we’re not saying that language models aren’t powerful. Yes, they can fine-tune text, spot clunky sentences, and help meet word limits. But they’re only as good as the instructions you give them.

If your command is “write my paper,” based on what? A data dump? A few bullet points? The synthesis AI produces may sound smooth, but will it ask the hard questions? Will it spot the methodological caveat that changes the interpretation? Will it draw the philosophical connections that underpin your field?

And this is a real issue. We’re already seeing more and more papers with excellent English but shaky structure, weak framing of the research question, and missing opportunities to interrogate the data in meaningful ways. Good grammar does not equal good science and it takes a lot of time and energy to unpick these types of issues and recreate a scientifically sound manuscript.

4. Our Approach at Insight Editing London

We’re committed to helping excellent science be published, not just look polished. Publishing is more than a final step—it’s the moment your work enters the global conversation and starts to shape the field.

But we don’t stop there. At Insight Editing London, we’re equally committed to teaching scientists—junior and senior alike—how to put together manuscripts that are clear, logical, and persuasive. Our goal is not only to strengthen a single paper, but to help you build lasting skills in scientific writing.

We do this by:

• Tracking our changes so you can see exactly how the text has been refined.
  
• Leaving constructive comments alongside the text, highlighting areas where logic, structure, or interpretation can be improved.
  
• Providing detailed editorial reports that explain, in a constructive way, how the manuscript could be enhanced further and how you can do it.

Moreover, our editorial team stays in direct touch with authors to answer queries and provide follow-up support as needed. This isn’t an extra—it’s part of our standard, human service.

So, if you’re working on a manuscript—whether for first submission or after reviewer feedback—and looking for some support, get in touch! We’ll help you ensure your paper is not just well written, but well argued, well structured, and ready to make an impact. And importantly, with a human touch!