Editing General Translation Writing

Now AI is here, do I really need a human translator or editor?

AI is conquering territories and taking few prisoners in tasks like translating, writing marketing and technical copy, creating images, planning and budgeting, and even writing novels. The IMF predicts that AI “will affect 40% of jobs around the world”. So who needs translators any more?

Can you finally ditch your translator, marketer, graphic designer or even the nice person who used to do all that general office admin?

Generative Artificial Intelligence (GAI, or AI) – that is, machine learning based on large language models (LLMs) and other training data like image databases – has made phenomenal leaps in the last year and is now replacing thousands of administrative and creative processes that have previously been performed by people. Last year’s Hollywood scriptwriters strike illustrated the extent to which GAI is already encroaching on jobs that used to be the exclusive realm of humans.

What does this mean for my business?

GAI is conquering territories and taking few prisoners in tasks like translating, writing marketing and technical copy, creating images, planning and budgeting, and even writing novels. As reported by The Guardian, the IMF has just released an analysis predicting that “about 60% of jobs in advanced economies such as the US and UK are exposed to AI and half of these jobs may be negatively affected”, while AI “will affect 40% of jobs around the world”.

An incorrectly computer-generated "blue dragon" showing the vagaries of AI creation.
GAI is prone to flights of fancy so you may need to specify to your GAI app exactly what a “dragon” looks like.

In the writing sphere, some independent authors are now using ChatGPT and other GAI apps for tasks such as planning novels, writing back cover blurbs and other marketing copy, or creating cover art. Consequently, Amazon now requires new publications to state whether they contain AI-generated content or not. In one AI writing group, an author boasted of having structured and written an entire novel in under three days using GAI. Another author claimed to have produced 30 novels in a year. This level of machine-assisted creative production ultimately begs the questions: “Do these novels sell and is the quality good enough for readers to buy them?” In the latter case, the author answered this by declaring a change of strategy: going forward, their intention is to focus on quality over quantity. However, it illustrates that we now truly inhabit an age in which a job like the one the character Julia had in George Orwell’s dystopian novel 1984 – working as a maintenance technician for the novel-writing machines – is no longer fiction.

Who needs translators any more?

In the translation sector, “traditional” computer-assisted translation (CAT) tools break down the translation task into sentence-length segments. These segments contain text in the original language (source) and the output language (target), and are stored in large databases. The translator can later retrieve and use these segments to maintain coherence and uniformity within a translation or across translations, as well as search for previously used terminology. Such databases are often shared by all employees within a company. In the case of the freelance translator, however, a single database is used to store each client’s translations in a particular language combination, meaning that they can search and reuse these translated segments where appropriate.

The new translation norm

With the advent of AI and LLM machine translation, translation segments are now sent into the cloud to be translated by a GAI app, and then are edited, checked and polished by a human translator. This is called post-editing, and it is fast becoming the norm in the translation sector.

A very British Aesop’s Tale

The recently reported British Post Office scandal should be held up as an Aesop’s tale on the pitfalls of blindly placing trust in IT. Nearly 1000 subpostmasters were falsely accused of theft, fraud and accounting crime due to faulty software. It only made the limelight thanks to a TV dramatisation starring Toby Bates, and required nothing short of an Act of Parliament to redress the wrongs. AI is thousands of times more complex than the IT software that ruined so many peoples’ lives for over two decades.

Can I trust machine translation, especially using AI?

Machine translation is often not as uniform as its makers would have us believe. The very fact of breaking a text down into segments means it may not translate so consistently, whereas it may pre-translate an entire text using the wrong terminology. It can choose the wrong word for a particular context, or translate an idiom literally in a way that means nothing in the target language. Furthermore, AI is wont to flights of fancy and inventing untruths. When it translates from a language like Spanish, which has a predilection for long chains of noun phrases, it tends to replicate that in English, despite the fact that we English speakers like to use a verb whenever possible so as to infuse our writing with vigour. The result is flat and boring. Neither does AI attempt to adapt culturally to its target audience. These are areas where a machine is still unable to best a human.

After dispensing with your graphic designer, remember to specify to your GAI app that humans generally have exactly four fingers and one thumb per hand.

Nevertheless, translation technologies – whether AI or earlier CAT iterations – vastly improve translators’ turnaround times and, when used correctly, can be a godsend in making terminology adjustments and ensuring uniformity across a single document or a client’s entire content. AI can be prompted for suggestions as to rewriting an unwieldy sentence or finding a fresh turn of phrase. When used correctly, the benefits for the translator – and by extension, their clients – are immense.

How can I ensure that my content stands out?

If you think that the Internet is already overflowing with superfluous content, things are set to expand exponentially. AI can churn out Internet content at a rate such that the sea in which we are currently swimming will expand into a veritable ocean. In this scenario, how can you ensure your content shines out more brightly? We have already become highly skilled at skimming over and scanning content on the Internet or in ebooks to extract just what we need.

Many readers of fiction, especially of the “whale genres” (readers of romance, crime and fantasy, who consume huge numbers of books) read using an extensive strategy, virtually skimming across the page to capture the story without taking much time over the details.

The human touch is key

As a writer, how do you snag your readers? You want them to pause on your page, read about your product, or take the time to read your book intensively. The key is the human touch. Compared to machines, humans are better at – and will continue to be better at for the foreseeable future – knowing how to create interest in a text. This is especially so for us experienced writers, translators and trained editors. We vary sentence length. Sometimes. With patient yet eloquent skill we convince, capture human flair, infuse our writing with brio to keep readers hooked. We are word professionals.

In translation, it is vitally important that you do not skip the post-editing process in a bid to cut costs or save time. Even better than using machine translation and then a post-editor is employing a translation professional from the start, one trained in using the appropriate technology in combination with their human expertise. As machine-produced writing becomes more ubiquitous, you need strategies to ensure that your article, book or marketing content rises head and shoulders above the rest of the field. The flat, anodyne register of much machine-written or machine-translated texts, whose style – or lack of – may be adequate or even appropriate for a handbook or technical guide, is not good enough to grab your potential readers’ or customers’ attention.

Make it human.

Google and other search engines boost high-quality content, regardless of whether it is human- or machine-created, but premium quality is what we humans still offer. So the next time you publish, keep a human in the loop. Investing in a human translator or editor will help ensure that your words go on to touch others in a truly human way.


Leave a Reply

Your email address will not be published. Required fields are marked *