{"id":2406,"date":"2025-10-17T13:24:25","date_gmt":"2025-10-17T13:24:25","guid":{"rendered":"https:\/\/pietervlugt.nl\/?page_id=2406"},"modified":"2025-10-27T09:45:26","modified_gmt":"2025-10-27T09:45:26","slug":"artikel-en","status":"publish","type":"page","link":"https:\/\/pietervlugt.nl\/?page_id=2406","title":{"rendered":"Post EN"},"content":{"rendered":"\t\t
\n\t\t\t\t
\n\t\t
\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t
\n\t\t\t\t\t\n\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\tNederlands<\/span>\n\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t
\n\t\t\t\t\t
\n\t\t
\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t

The ten problems with automated translation<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

AI technology has undeniably progressed strongly and currently shows a massive expansion. This wreaks havoc within the language industry and many agencies and translators are struggling to survive. Unfortunately it also leads to a deterioration in translation quality. Here I will describe a number of problems with AI and machine translations that, if not addressed by trained and experienced humans, will find their way onto platforms and websites and into marketing, documentation and apps.<\/p>

\u00a0<\/p>

1. Problems caused by a poor source quality.<\/strong><\/p>

A lot of documentation is written for the global market in English by local people who do not necessarily have a good grasp of that language. Moreover, technicians are usually not linguists and not used to explaining things in simple, clear language. At some point we have all seen incomprehensible or unintentionally funny manuals for products from non-English speaking countries. A human translator can often still come to a satisfying result by studying the images and drawings or asking the client for clarification. An AI or machine cannot take these additional steps and knows no doubt. When a source text is of poor quality or unclear, the automated translation may well turn out to be useless.<\/p>

\u00a0<\/p>

2. Mistranslation.<\/strong><\/p>

We have all seen this before. The machine or AI chooses the wrong solution for a word with several meanings.<\/p>

\u00a0<\/p>

3. Unnatural language.<\/strong><\/p>

Translation engines and AIs can by now produce grammatically correct sentences, but when source sentences deviate from established patterns, they quickly run into trouble. Their syntax will mirror the English too strongly, resulting in Dutch that seems off or is hard to read.<\/p>

\u00a0<\/p>

4. Problems around placeholders and tags.<\/strong><\/p>

Text in translation tools often contains tags that indicate where the text needs to go or that refer to chapters, paragraphs, pages, images, the table of contents or index. AIs and translation engines tend to struggle greatly with sentences that are interrupted by tags and translate the words between the tags literally and without any coherence. This renders titles, indexes, captions and references unusable.<\/p>

\u00a0<\/p>

5. Wrong terminology.<\/strong><\/p>

Many industries, organizations and companies have their own terminology. Translation engines and AI’s need to be specifically trained for the implementation of such terminology, otherwise they will simply use the terms that they know. Because training the technology is specialist work, automated translations will, in general, not contain the desired terms.<\/p>

\u00a0<\/p>

6. Non-translatable items are translated.<\/strong><\/p>

Some documents, especially manuals, contain sections that should not or only partly be translated. The software of displays springs to mind, but also button labels or specifications that contain imperial units. Translation engines and AIs do not make this distinction and produce translations for all of the content. Humans will have to rectify this during a post-edit, so that the content of the document reflects the actual product.<\/p>

\u00a0<\/p>

7. Punctuation problems.<\/strong><\/p>

Translation engines generally follow the punctuation of the source material. If the source has sloppy punctuation, the translation will be sloppy as well. In short, simple texts this can be remedied easily, but in large manuals full of specifications and numbers this quickly leads to problems. Some organizations and companies use specific punctuation rules that are included in a style guide. Automated translation systems must be trained accordingly, otherwise these rules will not be implemented.<\/p>

\u00a0<\/p>

8. Missed negations.<\/strong><\/p>

Translation engines and AIs regularly miss negating constructions in sentences, which turns the meaning of those sentences into their opposite. This can easily lead to dangerous situations, e.g. when working with or on machines or vehicles. Or consider the risks in medical, chemical or other sensitive environments.<\/p>

\u00a0<\/p>

9. Impoverishment of languages.<\/strong><\/p>

Translation engines and AIs greatly struggle with the recognition of style figures such as imagery, irony or hyperbole and usually translate these way too literal. Their translations of creative expressions like poems or lyrics are usually downright bad. Because the language of the machines is based on algorithms and the most common solutions are repeated, translations will become more uniform and smaller languages will look more and more like English. It appears that humans get used to this fast, but in the end languages will lose their expressive power, liveliness and unicity.<\/p>

\u00a0<\/p>

10. Plagiary.<\/strong><\/p>

The big players in the global field scrape the internet to collect source material for their large language models. They do this with little regard for copyrights. They consider the material to be theirs, unless the rights holder officially objects and prevails. This is not only unlawful, but brings with it the risk that texts written with the help of AI plagiarize pre-existing, copyrighted material.<\/p>

\u00a0<\/p>

There they are, the ten problems with automated translation. I do not purport that automated translation offers no value or should not be used. I do however argue for judicious application. Yes, some global players now have the tools and resources to produce usable AI translations within sharply defined projects. But even those texts still contain problems that can only be remedied by trained humans. Companies and organizations that use generally available tools will run into the problems described above. If they take their documentation and marketing seriously, they cannot do without trained and experienced language specialists.<\/p>

\u00a0<\/p>

In my daily practice I do not use an AI or translation engine, but I do work with so-called CAT tools (Computer Aided Translation) like Trados Studio and memoQ, or online applications like Smartling and XTM. These tools have a translation memory that stores earlier translations, so that I do not have to type relevant repetitions anew and the client does not have to pay for them again. To my mind, CAT tools are still the best and safest way to make translations, especially when they are combined with a project-specific term bank or glossary.<\/p>

\u00a0<\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t

\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

Pieter Vlugt, October 2025.<\/p>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"

Nederlands The ten problems with automated translation AI technology has undeniably progressed strongly and currently shows a massive expansion. This wreaks havoc within the language industry and many agencies and translators are struggling to survive. Unfortunately it also leads to a deterioration in translation quality. Here I will describe a number of problems with AI […]<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"elementor_header_footer","meta":{"site-sidebar-layout":"no-sidebar","site-content-layout":"page-builder","ast-site-content-layout":"full-width-container","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"disabled","ast-breadcrumbs-content":"","ast-featured-img":"disabled","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"enabled","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"class_list":["post-2406","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/pages\/2406","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2406"}],"version-history":[{"count":69,"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/pages\/2406\/revisions"}],"predecessor-version":[{"id":2624,"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=\/wp\/v2\/pages\/2406\/revisions\/2624"}],"wp:attachment":[{"href":"https:\/\/pietervlugt.nl\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2406"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}