Tina Julsgaard. Photo courtesy of FairLoc.

The translation industry has been rocked in recent years by developments in artificial intelligence technology. Machine translation tools like Google Translate and DeepL and generative chatbots like ChatGPT have advanced to such a degree that in November, one brazen CEO, keen to promote his startup's AI translation service, predicted that human translators might not even be needed three years from now.

While his claim raised many a skeptical eyebrow — there is quite a lot of unsubstantiated hype around the powers of AI — it is certainly true that these tools have shifted the landscape for professional translators.

Many companies requiring translation services have been wooed by the promise of faster results at a lower price, so previous translation opportunities on offer have disappeared or been replaced by lesser-paid post-editing jobs, which require linguists to review and improve a translation that is first done entirely by machine. One survey published in 2024 by the UK-based Society of Authors found that 36% of translators reported that they had lost work due to generative AI and 43% said they had lost income. Another survey from the same year by analyst firm Slator revealed that half of the freelance translators who participated had considered changing careers due to the impact of AI.

Though it's transforming the translation industry, the technology has a number of shortcomings, and such reliance on it over human expertise and imagination comes at a cost, says Tina Julsgaard, the managing director of a translation agency that specializes in Nordic languages.

The FairLoc logo.

“I’ve always seen translation as a highly creative process, a kind of craft, and allowing machines to do the heavy lifting will inevitably have an impact on quality,” she says. “Plus, who wants to live in a world of dull, uninspiring, automated content? It just seems like a shame to me.”

Julsgaard is also the co-creator of an initiative called FairLoc, a digital stamp launched in April 2024 that certifies a translation was created by humans, not machines. At Lingua, we're also advocates for human translation — for reasons both practical and related to our mission — so we reached out to Julsgaard to learn more about FairLoc and how it works.

Lingua: Tell us a little bit about your background in translation and localization.

Tina Julsgaard: I have always been fascinated by languages. I studied languages in high school and my major at university was Spanish. When I was 20, I went to Spain and fell in love with the culture and the way of life, not to mention my husband who is from Andalusia, the region of Spain where we live. I knew that I wanted to stay here and work with languages, and so that’s how my agency Comunica began.

Lingua: How did FairLoc come about? Is it the first of its kind?

Tina Julsgaard: The idea first came to me when I was at an industry conference a few years ago and I saw how the conversation had shifted towards a full-scale embrace of AI and of machine translation post-editing (MTPE). It felt a little disheartening; at times it almost sounded like my colleagues were starting to take it for granted that the role of human linguists would grow smaller and smaller as time went on. […]

As far as I know, FairLoc is the only certification scheme for human translations, although there are other initiatives calling for the same thing. For example, the Institute of Translation and Interpreting has published a Slow Translation Manifesto which advocates for the merits of a more considered translation process.

Lingua: How does FairLoc work in practice?

Tina Julsgaard: Translators and translation agencies can sign up as FairLoc Ambassadors. This allows them to issue stamps to clients for certain projects, provided they follow all the instructions in the writing agreement that they sign with FairLoc. This means that their translators can use AI and machine translation for inspiration, but not to create pre-translations that they then post-edit. Offering the stamp provides an extra incentive for clients to choose the human option, as it gives them another feather in their corporate social responsibility (CSR) cap. Ambassadors can also display the FairLoc logo on their website, and this can serve both as a door-opener to conversations about the human option, and as a signal to potential vendors that the agency values fairness and good conditions.

Our primary mechanism for ensuring compliance is the legal agreement – we think of it as being like an NDA. If you sign this agreement and then break it, you could risk legal consequences. On top of that, we also have a whistleblowing feature and a blacklist on our website. These tools help us to resolve any conflicts and provide an extra layer of security. In essence, they provide strong disincentives from braking the agreement and help protect the integrity of the FairLoc stamp.

Our hope is that, as the stamp spreads far and wide, it will encourage more conversations about what is happening in our industry, and help facilitate a shift back towards the human-led way of working that used to predominate in our sector.

“Bright ideas come from the heart. Inspiration can't be automised. It emerges organically from the human mind.” Illustration from the FairLoc website.

Lingua: Machine translation has been around for a while. Google Translate, for example, debuted in 2006. Why is now the moment for an initiative like FairLoc?

Tina Julsgaard: I think because the technology has developed so much since then, and in the last year – especially since the launch of ChatGPT and generative AI – the shift towards MTPE has really accelerated. More and more agencies are offering post-editing services, and clients are much more aware of them and the savings they offer both in terms of time and cost.

My worry was that we were at the beginning of a race to the bottom where suddenly we are willing to accept cuts in quality for the sake of lowering costs. I was concerned we were going to end up hollowing out our industry until what we’re left with is a tech sector rather than a creative, human sector.

It’s true that since then we’ve seen a rise in requests for purely human translations services at our agency, which is perhaps indicative of the fact that clients are noticing the drop in quality and the consequences this can bring. Even so, I still think there is a need for FairLoc, and that it will play a role in helping to balance things out again.

I also think that FairLoc can play a role in combatting an unethical practice that has begun to surface in our industry, where agencies sell clients human translation services but then use machine translation anyway. This is unfair on both the client and the translator. With FairLoc, this is not possible, and providers who do this may face legal consequences.

Lingua: Is it a good idea for someone to try to enter the industry as a translator or copywriter now in 2025? Is there enough room for humans?

Tina Julsgaard: This is such a difficult question to answer! I would never want to dissuade anyone from becoming a translator as I have such a huge love for this profession and the work that we do. It can be incredibly challenging, rewarding and enriching. Plus, the industry needs talented and creative humans who can think outside the box and find innovative solutions to tricky problems.

At the same time, though, it would be remiss of me not to mention that lots of linguists are leaving the industry, including some of the most talented. They quite simply feel that their work is no longer valued the way that it used to be, and they aren’t able to earn enough money to keep going. This is definitely something that graduates today should consider before deciding to train as a translator or a copywriter.

I do believe that things will change in the future, quite simply because we need the quality that human creativity brings to the page. My hope is that as clients become more and more aware of the effect that this drop in quality produces, the more they will realise that not only is there room for humans in this industry, they are needed. And hopefully FairLoc can play a role in speeding up that process of realisation.

Lingua: Do you think there will always be demand for human-driven translation? Why or why not? Are there analogous experiences in other industries that we can look towards?

Tina Julsgaard: One thing I’ve noticed recently is a kind of backlash online against AI-generated images. It seems that at first, there was a novelty factor to these kinds of pictures, and people thought they were cool. As they become more common, however, they are starting to become synonymous with laziness and cheapness. They are a bad look for companies as they give the impression that the company doesn’t care, or that they don’t have the money to invest in proper design.

It is for the same reason that I think there will always be a need for human-driven translation. People are becoming better and better at recognising the tell-tale signs of AI and machine translation, and this will prove a turn-off, ultimately impeding the ability of companies to connect with customers.

Within less creative segments, it may be the case that human-driven translations become less and less desirable. I’m thinking, for example, of formulaic policies and legal texts that follow predictable patterns. However, there will always be a need for humans in some part of the process – there are very few people out there foolish enough to entrust a legal or medical text to machine translation without any human oversight. The risks of this are quite simply too great.

A still from a FairLoc promotional video.

Lingua: Have you noticed any trends in the kinds of businesses that seek out human-driven translation vs. the kinds of businesses that settle for pure machine translation?

Tina Julsgaard: I think the general rule is that clients prefer human translation when they prioritise their reputation and brand image not just in their home market, but abroad as well. Clients ordering translations of marketing texts, website texts and other types of visible, high-profile copy tend to understand that these texts require human translation in order to retain their hook and appeal.

Lingua: FairLoc also stands for fair practices in the translation industry. Could you unpack that a little?

Tina Julsgaard: Absolutely. If you’re on LinkedIn, it’s hard not to notice that many translators are feeling unfairly treated at the minute. After years of hard and dedicated work, AI and machine translation are being used as cudgels to beat down rates and conditions. Some linguists have seen their work dry up completely, while others are having to accept lower rates for a more frustrating and less rewarding form of what they used to do so well.

What makes this all the worse is that machine translation is only possible because of their work in the first place. It has been trained on historical translations, but the translators aren’t receiving compensation or recognition for their role in this. Their data has been taken and turned against them, and that is quite simply unfair, no matter how you look at it.

I know that many businesses today are concerned with their ethics and corporate social responsibility, and I’m sure they don’t want to be part of something that treats hard working professionals in this way. The problem is, they don’t really know what is going on behind the scenes, and so that is one thing FairLoc aims to do. We want to explain to clients that vendors in their supply chain are being treated unfairly, and we want to give them a way to support them and help change the situation.

Lingua: When it comes to a translator’s personal practice, what does a healthy relationship with machine translation tools look like, in your opinion?

Tina Julsgaard: We advocate for a human-led or human-driven approach. We’re not at all anti-machine translation or anti-technology, but we think automation tools should be used as aids to assist and inspire us. At the minute, we’ve got things a little back-to-front. We’re letting the machines do the heavy lifting by creating a pre-translation, and then we’re doing the dull, mechanical part of the work when we go in and painstakingly tweak and review the machine output.

Instead, we think humans should be doing the actual creation work, while using machine translation and AI as sources of inspiration in this process. The best tools, for example, are the ones that include a machine translation suggestion to the side of the target column. Something that translators can glance at, change and build upon, rather than something that constrains them.

Lingua: As a translator, what do you personally enjoy most about the art of translation?

Tina Julsgaard: I love the satisfaction of solving a tricky problem. Like a bit of wordplay that you think just can’t be matched in the target language, until suddenly you get that lightbulb moment and the perfect phrasing comes to mind. I think translation is a uniquely challenging discipline in that it requires a lot of creativity and free thinking, but at the same time, you have all these other requirements to consider – you always need to match the meaning, the tone and the intention of the original. It’s like a form of puzzle-solving, and it’s incredibly satisfying when you figure out how to get it right.

Lingua: What is the future for FairLoc?

Tina Julsgaard: We are currently exploring the possibility of working together with computer-assisted translation (CAT) tools to create a FairLoc set-up for linguists that ensures machine translation and AI are used in a compatible way. This could even allow us to issue stamps automatically through the CAT tools and thus help us solidify FairLoc as a service that clients can order. So far, the response to FairLoc has been very positive and we believe that it will only continue to grow in the months and years ahead.