In our previous blog post, we explored how artificial intelligence (AI) is transforming publishing and why human expertise remains essential to professional copy-editing. AI can now check grammar, improve sentence structure, summarise documents and suggest alternative wording in seconds. For many, it has become a valuable addition to the editorial toolkit.
Building on that discussion, we now look more closely at where the limitations of AI-assisted editing become apparent in practice and why the judgement of an experienced copy-editor continues to matter.
Professional copy-editing has always involved far more than correcting spelling, punctuation or grammar. It requires editors to understand the author’s intentions, recognise the client’s objectives and ensure that readers receive content that is clear and fit for purpose. AI can support many aspects of that process, but professional editors remain responsible for making the decisions that balance these sometimes competing priorities.
The key difference here is the ability to exercise judgement. Every editorial decision requires someone to consider not only whether the text is technically correct, but also whether it communicates effectively for its intended audience and purpose.
Every sentence prompts editorial questions
When an experienced copy-editor reads a document, they are constantly asking questions.
- Does the text actually say what the author means?
- Does the text reflect the tone the client is looking for?
- Would restructuring sections of text improve the flow or strengthen the argument/narrative?
- Is the content accessible, inclusive and suitable for its intended audience?
- Does the text contradict something stated earlier?
- Could this wording unintentionally mislead readers?
- Is anything important missing?
These are not grammatical questions. They are editorial ones.
Answering such questions requires an understanding of the author’s intent, the client’s objectives and the reader’s perspective. AI can identify patterns, highlight potential issues and offer valuable suggestions, but experienced editors remain responsible for deciding whether those suggestions are appropriate for the document at hand.
Reading for meaning, not simply for words
One of the biggest misconceptions about AI-assisted editing is that, if every word is correctly spelt and the grammar is sound, then the sentence must also be correct.
Experienced editors recognise that this is not always the case.
Our own testing and evaluation of AI-assisted editing, alongside our experience of working with AI-generated content, have highlighted some of the limitations of the technology. The following examples are based on recent issues we have encountered during our own editorial work on AI-generated content. They illustrate that AI does not always recognise when a word or sentence fails to convey the author’s intended meaning.
Consider the sentence:
It can be interred from the results that the intervention improved patient outcomes.
Every word is correctly spelt. The grammar is acceptable. Yet the sentence is meaningless because interred is not the right word. This is where professional editorial judgement comes in. We recognise that what would actually make sense here is:
It can be inferred from the results that the intervention improved patient outcomes.
Another recent example we encountered involved ‘incarnation’ being used instead of ‘incarceration’. Both are perfectly valid English words, yet only one had the correct meaning for the sentence. Once again, the problem wasn’t spelling or grammar; it was context.
Professional editors do far more than correct individual words. They ensure that the author’s intended meaning is communicated clearly and consistently from beginning to end.
Clarity matters just as much as correctness
Sometimes a sentence is technically correct but still does not communicate as clearly as it could.
For example:
The treatment significantly reduced patient mortality compared with previous studies.
At first glance, the sentence appears acceptable. On closer inspection, however, the comparison is unclear.
If the comparison is with mortality rates reported in previous studies:
The mortality rate among patients receiving this treatment was lower than the rates reported in previous studies, regardless of which treatments were used.
If the comparison is with other treatments evaluated in previous studies:
The treatment reduced patient mortality more than the other treatments evaluated in previous studies.
In each case, the intended comparison is made explicit.
However, the two interpretations are not necessarily mutually exclusive. A treatment could produce a lower mortality rate than those reported in previous studies without necessarily being more effective than the treatments evaluated in those studies.
Again, the issue is not grammar but interpretation. An experienced editor asks whether the sentence says exactly what the author intended and whether readers are likely to understand it in the same way. Where the intended meaning is uncertain, an author query may be needed rather than an editorial assumption.
Knowing when not to edit
Perhaps the most underrated editorial skill is knowing when to leave things unchanged.
Every author has a voice. Every organisation has its own style. Every publication has a purpose.
An experienced copy-editor understands that good editing should improve a piece of writing without removing its personality or changing its intent.
Sometimes the right editorial decision is to preserve an unusual turn of phrase because it captures the author’s voice. At other times, it is to retain specialist terminology because precision is more important than simplicity.
AI-generated suggestions may appear grammatically correct or stylistically polished, but that does not necessarily mean they improve the text. Recognising when a suggested change would alter the intended meaning or diminish the author’s voice is an important part of professional editorial judgement.
Editorial judgement protects quality and trust
Publishers, charities, businesses and public bodies all have reputations to protect. Their readers expect information that is trustworthy. Editorial judgement plays an important role in safeguarding that trust by identifying ambiguity, questioning inconsistencies and helping ensure that published content communicates exactly what it is intended to.
For many organisations, that confidence in the editorial process is every bit as important as efficiency.
AI is changing editorial workflows
As we discussed here, AI is already changing how editorial teams work and becoming an increasingly valuable tool. By reducing the time spent on routine tasks where appropriate, AI can allow editors to devote more attention to the work that benefits most from human expertise: improving clarity, strengthening structure, questioning inconsistencies, applying house style, preserving author voice and ensuring that content is clear and appropriate for its audience.
In many respects, AI is making the value of experienced editors more visible rather than less.
At the same time, we recognise that AI is not the right tool for every project. Many of our clients do not permit the use of AI when working on their documents, while others have specific requirements governing its use. Some clients prefer (or require) a fully human editorial process because of organisational policy, confidentiality, regulatory requirements or personal preference. Our role is to understand each client’s requirements and tailor our editorial approach accordingly.
At Prepress Projects, we are actively exploring what developments in AI mean for professional editing and for our business. Our testing helps us understand where the technology may offer genuine benefits, where its limitations remain and how editorial workflows might evolve. However, any potential use of AI must be considered in the context of our clients’ requirements and our commitment to maintaining high editorial standards.
The future is collaboration
Publishing has always evolved alongside technology.
From hot-metal typesetting to desktop publishing, from paper proofs to digital workflows, each innovation has changed the way publishers work.
AI is simply the latest chapter in that story.
The future of publishing is not about choosing between AI and human expertise. Technology can assist with many editorial tasks, but professional judgement is still essential; the key is understanding where each delivers the greatest value.
For Prepress Projects, this means continuing to evaluate the opportunities presented by AI while recognising that its suitability will depend on the nature of the work and, importantly, the requirements of our clients.
In our next blog in this series, we will explore how clients are responding to AI, from their approaches to AI-generated content and its use in editorial workflows to the requirements increasingly emerging in tendering and procurement processes.