From the boardrooms of Silicon Valley to the bustling hubs of Asia, language industry leaders recognized the transformative role of AI in translation, interpreting, and multilingual content generation in 2024, particularly the versatility of large language models (LLMs) and the fundamental shift in how laypeople can now interact with language itself.
The AI adoption and adaptation journey is not without its challenges. Experts on the technology side highlight the rigorous demands of enterprise-level AI implementation while voices in management, ops, and linguistics emphasize that human creativity and nuance remain essential and irreplaceable.
The following collection of insights paints a picture of an industry still navigating exciting possibilities and inevitable complexities as AI is integrated into more processes.
January 2024. During SlatorPod episode #198, John Tinsley, VP of AI Solutions at Translated said “The second ‘L’ in LLM is language. When we talk about transformers in neural machine translation, the ’T’ in GPT stands for transformer. All these technologies started with the problem of language, and we’re in a super exciting time about working out how we can use it for ourselves, for customers and it’s really exciting, and I would hate not to be involved in it.”
February 2024. Bart Maczynski, VP of Machine Learning at Language Weaver said during SlatorPod episode #199 that “Enterprise MT has to survive the scrutiny of a sophisticated enterprise or government buyer, and that means multiple aspects like data security, scalability, ease of integration, adaptability, user experience, and things like risk mitigation, or even acceptable licensing models. Now the trick here is that the solution has to tick all of the boxes all the time, not some of the boxes some of the time.”
March 2024. Chris Reynolds, EVP and GM of Worldwide Localization and Fulfillment, Deluxe said on SlatorPod episode #205: “I think the creative aspects of dubbing shouldn’t be underestimated. For accessibility, maybe opening up new languages that aren’t usually dubbed and doing voiceover dubs, or just a lower quality dub … I can see people testing that, but I don’t see it replacing human voice actors. There’s an emotional performance there … Performance nuance is important.”
March 2024. On the topic of leveraging resources for AI, Gabriel Fairman, CEO at Bureau Works stated during SlatorCon Remote: “So if you have any deviations, for instance, let’s say your TM is full of gender bias … all your feeds are going to be full of gender bias, and it’s going to be hard to steer it … if your entire foundation is driving it in another direction.”
April 2024. On SlatorPod episode #207, Dr. B.J. Woodstein, Professor, Translator, and Writer commented that “JK Rowling, the author of the Harry Potter books, was infamous for not wanting to speak to her translators. I don’t know what that was about and It seems to me really short-sighted … That was a case where it was all mediated by somebody else. I think my understanding was that lots of translators got a list from her publisher of things they were expected to do … without showing any knowledge of the actual languages in question.”
April 2024. Daniel Wilson, Founder and CEO, XRI Global stated on SlatorPod episode #209 about his firm’s work: “What we’re able to do is take one of these smaller LLMs and the ASR, TTS, and MT, we stitch the whole thing together, and it runs on a single laptop with a single GPU. Even for offline communities or communities where the internet is too expensive, they can actually interact with, do conversational AI, and chat with resources that haven’t been translated.”
May 2024. On the topic of human expertise, Richard Parnell, General Manager of Linguamatics at IQVIA said during SlatorCon London that “We’re finding that that’s really important. Having that combination of [language industry and life sciences] subject matter experts is working very well and it’s enabling us to be able to do things that I think we otherwise probably wouldn’t think about or challenge ourselves to do.”
May 2024. Charles Campbell, President of tbo. commented about automation during SlatorPod episode #210: “I think that automation today is more of a discussion about how automated you are rather than whether or not you are automated or not. If you’re not automated in any way, then you have a choice, which is, I think, to start automating to some degree or another without losing the essence of your value and your core business, without losing the human touch or gradually wither on the vine.”
June 2024. Georg Ell, CEO at Phrase said during SlatorPod episode #213 that “Historically, you might have had a chance to do some degree of quality scoring early on in that process. Now you have no chance because the machine can generate it so fast that it’s all going out the door very, very quickly. This goes to the hyper-automation, hyper-scale type journey we’ve been talking about, where I think volumes are going to go through the roof as more and more generative AI, more and more multilingual content at source is used by enterprises to generate content in real time.”
June 2024. During SlatorPod episode #214, Marina Ilari, CEO, Terra Translations and Terra Localizations “Game localization is very culture-sensitive. It is creative in nature and it really seeks to resonate with the players, with the gamers … I think AI, particularly LLMs, can help us in many different processes … create efficiencies to advance tasks that we had no resources, no time, no budget to do before.”
Slator Translation as a Feature (TaaF) Report
The Slator Translation as a Feature (TaaF) Report is a vital and concise guide on how AI translation is becoming an integral feature in enterprise technology.
June 2024. About the use of retrieval augmented generation (RAG), Roeland Hofkens, Chief Product and Technology Officer at LanguageWire said during SlatorCon Remote that “We do not hold all the information of our customers. We have what they send to us for translation, and that’s what’s in the TMS in the end. So it’s a very nice multilingual data set, but it does not cover everything.”
June 2024. Discussing use-case-driven automated QA during SlatorCon Remote, Simone Bohnenberger-Rich, PhD, Chief Product Office at Phrase commented “Customers can determine, depending on the use case and the need, how far they want to move the needle towards speed, scalability, and automation, and for what type of assets they prefer an exception loop.”
August 2024. On SlatorPod episode #221, Spence Green, CEO and Co-founder of LILT said “I think for the subset of use cases that really matter in enterprise localization, which are translation, quality control like QA, rewriting, and summarization, those four use cases, you don’t need a trillion parameters to solve those use cases. If you use smaller models, you get better performance cost.”
September 2024. Speaking about AI as an opportunity during SlatorCon Silicon Valley, Andrew Doane, Senior Vice President, K1 Investment Management said during SlatorCon Silicon Valley “What we found, going back a year ago, was that there were some companies that jumped right in and maybe over-promised and under-delivered a little bit, and maybe got out over their skis… There’s a long way to go before there’s no longer a defined human element in the translation process.”
September 2024. On SlatorPod episode #223, Christopher Kurz, Head of Translation Management at ENERCON, commented about ISO 5050 that “You can evaluate everything … It doesn’t matter if it’s raw MT output, post-edited MT output, if it’s human translation only, if it’s human translation memory. This is the basic idea of 5060. It always refers to the specifications even if MT has ticked off all the specifications. You can evaluate it, but your evaluation has to be done by a human.”
2024 Slator Pro Guide: Translation AI
The 2024 Slator Pro Guide presents 20 new and impactful ways that LLMs can be used to enhance translation workflows.
October 2024. Sijie Wei, Co-CEO of EC Innovations, said during SlatorPod episode #229 that “I read that APAC represents around 20% of the globalization services revenue globally. While APAC has become a very big export market from a macroeconomics point of view, the localization maturity is still catching up to the speed of the growth of the markets overall.”
November 2024. During the Translating Europe Forum, keynote speaker Marina Pantcheva, Director of Linguistic AI Services at RWS Group stated that “LLMs just average the language choices that other writers have made before them, so basically they’re going for the least interesting choice … They lack persistent memory so they do not remember how they responded to your prompt yesterday.”
November 2024. On SlatorPod episode #230, Christian Elongue, Managing Director of Kabod Group, commented that “In general, there is a need of increasing ownership from African partners, African organizations, in sponsoring the development of language technology solutions and not to let this only be driven or mainly be driven by Google, by Mozilla or Meta. I think there’s so much more that can be done if Africans also enter that space.”
December 2024. Speaking on SlatorPod episode #235 about AI in interpreting, Katharine Allen of the SAFE AI Task Force said “This technology is going to do great good and it has potential to do great harm. We’d like to put some basic consumer protections the same way you would with your car … some kind of guideline, some guidance, and some kind of guardrails around what criteria to apply … ‘when is it all human, when is it all AI, when is it more likely to be some kind of hybrid effort?’”
December 2024. During a public inquiry session about AI interpreting in the UK courts, Lord Porter of Spalding stated “Surely if the health service is starting to embrace this technology, your industry must embrace it. From a consumer’s perspective, yours is a pretty safe place to put it in. It is just language. The consequences are not as dire as they would be in the health service. … If you do not engage with it, you will be so far behind the curve that you will always be playing catch-up with it.”
