Two translators with the same assignment got very different answers.
A translation firm offered both professionals the same project—a large database of food product names and culinary labels to be translated into four languages. One turned it down, saying the research demands made quality work impossible at the rate offered. The other accepted, delivered a polished result, and the client approved it without a single revision.
A recent piece from the GTS Translation Blog uses that real-life situation to ask a question the whole industry is wrestling with: what has AI actually changed about professional translation work? For Fr. Philip Johnson, a French translator and academic researcher at Hofstra University, this isn’t a hypothetical debate. It’s the day-to-day reality of his profession.
Research Is Faster
Translation used to require hours of manual research before a single word got written. Specialized terminology, regional variants, labeling conventions—tracking all of it down was genuinely time-consuming work.
Today, AI tools, bilingual databases, and search resources have made the lookup process much faster. The bottleneck has shifted from discovering terminology to applying judgment—choosing the right equivalent, maintaining consistency, matching the appropriate tone, and making dozens of small decisions that shape a finished translation.
What AI Can and Can’t Do
AI is genuinely useful for locating terminology, checking real-world usage, and scanning parallel references. What it cannot do is understand cultural context, determine whether a word fits its specific purpose, recognize differences in register and inflection, uphold consistency across a large project, or take professional responsibility for the final result.
That last point matters more than it might seem. When a translation goes out into the world—on a product label, in a legal document, in an academic publication—a professional made deliberate choices to get it there. AI doesn’t carry that accountability.
This is especially relevant in Fr. Philip Johnson’s work. Academic translation isn’t only about finding equivalent words. It’s about correctly representing a scholar’s meaning across languages. A mistranslation in a research context doesn’t just read awkwardly—it can distort an argument or introduce errors that flow through other work.
A Shift, Not a Replacement
The GTS piece makes a good point that productivity improvements don’t eliminate expertise, but they can elevate it. AI rewards translators with real subject-matter knowledge, reduces time spent on repetitive research, and shifts the value of the work toward judgment, reliability, and consistency rather than hours logged in terminology lookups.
The profession is changing. But the core of what makes a translator valuable—cultural knowledge, linguistic precision, professional accountability—isn’t something a tool can replicate.
The Bottom Line
AI has made parts of translation work faster and less tedious. That’s genuinely useful. But faster research doesn’t produce better translations on its own. Someone still has to understand what the words mean, why they matter, and whether the final result actually does its job.
That part has always required a human. It still does.
Leave a Reply