Artificial Intelligence & Human Endeavour
Part I of III: AI, Translation Services, and Cultural (out)Reach
Recently, Fiona Steven opened a conversation on LinkedIn with “How do you make a brand feel at home in different languages and cultures?”. As part of exploring the answer to that question, she shared some insights from her own background and experiences living and working between Mandarin, English and Malaysian languages, with their associated colloquial, idiomatic and cultural differences. She was essentially making a [compelling] case that translators provide more than words, they provide resonance and reach.
Similarly, and in the same recent timeframe, many others have shared frustrations as professionals providing translation services. They feel undervalued, underpaid, and under threat from AI (in particular). Many of these people have advocated that the majority simply don’t understand the time, effort, and worth that good translators bring to their work in the service of their clients. Such work can make a client’s goods and services greater than the sum of their parts, so it carries a greater monetary cost reflective of that expertise, care, and time, which also provide intangible benefits to their clients. We agree.
It is probably of no comfort, and even less of a surprise, for translation professionals to hear that all employment sectors have similar problems with niche and/or highly specialized expertise. Statisticians take a dim view of biomedical scientists that turn up asking “can you power this study for me? By the way the grant is due/regulatory filing deadline is tomorrow”, with nary an offer of compensation in sight, and sometimes even lacking basic please-and-thank-you. But it happens all-too-often. Similarly, government and other elected officials regularly “need” reports, reviews, policies, procedures and other “basic” documents from Civil Servants. Yesterday.
So, what gives?
Is it due to inherent misunderstanding or mistrust of expertise?
Has the real cost (time, monetary) of translation services outpaced their perceived value?
Are increasing complexities causing people to seek the simplest solutions?
Is the pace of modern life overwhelming?
Perhaps it’s the rise of the internet and other tech enabling thought processes such as
“I don’t need a statistician because I can find the equation for propensity scores”,
“There’s an app for that”, or
“I don’t need ‘experts’, because I can do my research on the internet.”
We think it’s all of the above.
In Douglas Adams’ Hitch Hiker’s Guide to the Galaxy a universal translator called a Babel Fish (the name itself has deeper translational meaning) is placed in the ear of an individual to translate perfectly, in real-time. This is a modern version of innumerable fantasies created over the millennia as humans strive for better understanding of each other, beginning [but not ending] with language comprehension. And literal translation alone is almost never enough because tone, context and cultural references are critical, to understanding. Even people that nominally speak the same language can misunderstand or misinterpret each other when colloquialisms and cultural references are included in conversation, just ask a Californian having a discussion with someone from Tennessee (and vice versa) or an Englishman in Australia. To Mr. Adams’ credit, he envisaged a sentient being as a translator with a symbiotic relationship where language was translated idiomatically, with meaning and context. At present, AI does not or cannot provide such, and this is one of the human advantages.
AI is entirely derivative, and it's not judiciously derivative. Unlike, for example, a student having AI write an entire assignment who then passes an eye over the assignment to determine the writing’s relevance, and clean up anything that is outright ludicrous. In this example, the student commonly doesn’t have the experience to recognize if "facts" and statements are wrong or even mildly silly. So, in addition to lacking the ability to sift correct from incorrect information, one could argue that current AI programs lack higher reasoning, what we might call "judgement". In the case of the student, judgement exists but is incomplete due to lack of experience. The major difference is that "judgement" is or should be provided by a teacher or professor [pun definitely intended] who uses prior experience and knowledge to recognize error organically. So again, the current human advantage is and remains, reasoning and judgement that occurs spontaneously and creatively. Put another way, the suggestion that AI could meaningfully translate an epic poem such as Homer’s Odyssey or great and complex works of literature such as War and Peace (Tolstoy, obv.) is, currently; ludicrous. An AI-generated translation would lack much of what makes these works great and good – their humanity.
Beyond the perspective of human skillsets, there are also economic arguments to the effect that the speed and almost zero cost of using AI to translate for free something small and straight-forward like a 400-word flyer, outweigh the small economic loss to translators. By their nature flyers are supposed to be direct, simple and without ambiguity, meaning that AI’s lack of context and subtlety may be less concerning. Depending on your perspective, this also may be less harmful because translation tasks of this type are inexpensive in comparison to the time spent contacting individuals or companies, getting quotes and receiving the work. Depending on the language the content is prepared in and/or translated to; it’s about 25 – 65 cents Canadian per word for human translation services right now, so $100 - $260 plus taxes and fees for that 400-word flyer. For most businesses or public institutions these costs are very affordable, or next-to-nothing. The flip side of that argument is that fast, affordable translation may be low cost compared to administrative costs/time and lead time incurred, but can account for a translator’s bread-and-butter thorough higher volume of work performed. Secondly, despite what many people think, AI does indeed have a cost, and we may yet find it crippling. AI did not appear from nowhere and is not sustained by nothing. Vast data centers power AI and these require space, and energy to generate responses, graphics, videos and other output. It was reported as recently as July 2025 that ChatGPT received 2.5 billion prompts daily, and that Google received roughly 5 trillion queries per year (this is about 14 billion a day). The enormous scale of these online interactions requires equivalent energy, water and other resources, of many small countries. So no, AI is not “free”.
We’ll leave it there for now, because in the second and third part of this short series we intend to cover economic issues more completely, along with mass communication.
In the meantime, to stop the slop: keep creating and shine!