Episode Transcript
[00:00:03] Speaker A: Today we have a conversation with Marina Hancheva, Belen Aguyo Garcia about translation, about AI, about the future of the language services industry and everything in between.
Hello and welcome to Localization Today. My name is Eddie Arrieta, CEO here at Multilingual Media.
[00:00:35] Speaker B: So great to be here. So much fun. Thank you, Eddie.
[00:00:38] Speaker A: Yeah, this is a lot of fun. And we've had some pending conversations from the past. So, you know, before off the mic, I was mentioning to Marina and Belen. We should never assume that everyone knows you. And I know some people in the industry already know you. Belen. Marina, would you mind introducing yourselves? Marina. And then with Belen.
[00:01:03] Speaker C: So my name is Marina Pancheva and I am currently linguistic. No, director of Linguistic AI Services at rws. I'm leading a division which implements AI technology in our services in what I would like to say a meaningful way so that it's really useful for translators and for clients. But that's kind of a new job that I have. I started my career as a translator, as most of us did. I was also a very passionate L2 teacher. So I was teaching all sorts of languages. And I've always been very interested in languages. So maybe that's the reason I went to do PhD in theoretical linguistics. I never really officially studied translation, but I did theoretical linguistics. And then life took me to another country where I had to find a job, and I decided to move to the language industry and see what that was like. And I loved it. I loved the space speed, I love the little element of risk and not knowing what's going to happen tomorrow. And probably this love for risk and the unknown also got me on the AI train, because when ChatGPT exploded in November 2022, I found myself very much attracted to exploring the capabilities of AI, finding out what that actually was. And a few months later, I found myself leading a whole division doing just that.
[00:02:34] Speaker A: That is. That is fantastic. Thank you so much. And we'll delve deeper into that for sure. Belen. Who is Belen?
[00:02:43] Speaker B: Hey.
Hi, I'm Belen. And, well, I have to say I'm always fascinated at Marina's stories about her life. And before I introduce myself, I want to say that she's an inspiration to me to get better, a better professional, you know, be more intellectual. And also thank you, Marina, for inspiring me. And yes, I've been in the industry for almost 15 years. Purely always the same, the path of translation. So I studied translation and interpreting. I am from Spain, I live in Spain, studied translation and interpreting here, because when I was in high school, I loved English. That was the only foreign language I had access to back then. And I knew I wanted to do something with languages. But teaching, maybe like teaching languages was not something that was very appealing to me back then. And I had this booklet with all the possible BAs and degrees I could take at the university, and translation was one of them. And I thought, well, this might be something I can do.
So that's where it all started then. Most of my career I've been working at the service provider side, especially in gaming and multimedia and entertainment. So I've always been working in the more creative spaces of localization, which is really fun and cool.
In the meantime, I also tried my luck in academia and research. So I also pursued a PhD in, in translation studies. Specifically, my PhD was about how to integrate accessibility, so captions and audio description and sign language in immersive content, in virtual reality content. And I focused on captions and subtitling and it was such a fun project also funded by the European Union. And in that period I also got to learn more about accessibility, access services and how to work with end users, the receptors of accessibility. And it was super inspiring.
So that's something. It's such a great experience that I bring with me. I've also been a researcher at nimsi, Eddie, you know it very well, Nimzi. And I have super great memories working with Nimzi. And still I am collaborating with them in some research, which is. Yeah, so basically I love researching, I love localization, I love sharing what I know. And most recently my current role at Terra Localizations, which is an LSP also focused on game localization and also other verticals. But gaming is our biggest one here. I am executive consultant of innovation. So what I'm doing here is driving the innovation program, which of course includes technology, of course includes AI, because AI is the big thing about the future and how LSPs will transform or evolve in the age of AI. So basically trying to find strategies, trying to find ways where my company and the industry as a whole can keep thriving despite the threats of AI and technology and developers.
[00:06:09] Speaker A: And that is the main conversation. Thank you, Marina and Belen, for introducing yourselves. And we are forced to talk about artificial intelligence. And you are AI strategies, AI innovation leaders.
How was that journey for you going from being more on the translator side or any side that you were thinking about and moving to become an AI thought leader. And that's difficult to say today what type of AI thought leader you are. So could you, could you tell me your thoughts on that and anyone would.
[00:06:46] Speaker C: Do so maybe I will start. And I actually want to go back a little bit and say one more thing that Belen and I have in common. Purely professionally. Before we became AI thought leaders, as you call it, we were both involved in a project that involved working with large translator community. So first when we started talking, it was about community management and freelancers. And then we met at a conference to find out that now we are both doing AI. So it was very interesting to find out that we were both attracted to the same topic, AI. And we continue talking to each other, but this time about a different thing.
So back to your question then, Eddie. You know, I think I've always been, I wouldn't call it a thought leader, but somebody who wanted to try out something new. This is probably also the reason I was doing research, because that is the very essence of research. You have a problem, nobody knows the answer to that problem. It is a totally new thing. And you need to go out there in this completely uncharted territory. Okay, you do have some marks, right? You do have previous research or some articles to base yourself on, but you don't know whether you will even find the answer to your questions. Maybe they don't exist yet. Maybe we are not that advanced to find them. And I applied that strategy to my research.
I also applied it to what I was doing before I got involved with AI.
Back then I was leading a team that developed solutions for crowd localization. So I had the freedom to come up with new ideas and actually implement them, experiment a lot and try out new things. So when AI appeared in the form of large language models that were accessible to the public and anybody could play with them and they were powerful enough already to produce good results, I started looking into that simply because that was another new thing that could be implemented. I was curious and naturally attracted to finding out what their capabilities were.
And very soon after that, it became clear to me that this knowledge is going to be mandatory. It is inevitable. Clients interest increased so steeply within the first, I would say couple of weeks even. So by February 2023, which was just three months after GPT 3.5 was released, clients were asking, you know, what can we do with AI? We want to do something with AI.
So they showed me that, you know, this is the way to go. This is the new territory that needs to be explored. So I just started doing that more and more until it became my full time job.
[00:09:35] Speaker B: And my story is a little bit different how I ended up doing what I do So I always loved language technology since the beginning of time. I remember when I started working in the industry, we were using workbench to localize games, which was such a nightmare, honestly, back then, because we had to copy paste things in word and then go back to the Excel sheet and yeah, not fun times for localization back then. And then my company acquired memoq, which was like a super innovative tool back then, especially for game localization. And I started using MemoQ and defining the workflows, understanding how we could leverage it, how we could automate more of the processes, more on the PM side and so on and so forth. So that's always been something that I've been attracted to since I started in this industry.
Actually, I worked at Fun Fact, I worked at memoq for almost a year and I loved working with a product team.
The intersection between product sales, the end users and all is such a fascinating topic for me. So product technology always been such a. Yes, that's a great topic that I enjoy working on. How I ended up doing AI was more practical because in my previous company I used to work at Deluxe, which is another service provider, but more aimed at media.
So working with Netflix, Amazon, Disney, Warner and all these big production companies and they decided to invest in an AI company called uptech. And you know them as well.
So the idea. So when this project started, I was assigned as AI strategist for it. So basically trying to integrate that technology with the workloads of the company, how we could improve the technology for the use cases that we had, how we could design a service offering that would be appealing for the type of customers that we had with the high expectations. And that included all sorts of things. We were working with neural machine translation, but also automatic speech recognition for captions and also synthetic voices. So honestly, I was very excited with this project because it meant that I could learn so much new stuff that I didn't know before. And what great thing about it as well is that I got to work with the scientists, with the UPTech team, which was such a great experience for me because I learned a lot. And also usually when you're at an LSP and you, for example, use DeepL or Amazon MT or Google MT, and you want to implement that in your workflows and so on and so forth, there is as much that you can do with that. You cannot change the foundational engines and the language models. But working with AppTech, that was possible because they were the scientists designing the actual engines. And that was, yeah, it was fascinating to learn how it worked and be able to influence the technology so that it would become better for translators. And, yeah, that's how I ended up being in AI.
[00:13:13] Speaker A: And it's very exciting to see different paths, different career paths and how you get to the expertise. And I know we're going to be talking about the future and kind of like what you see there, which I know it's going to be very interesting for our audience. But before we get there, I'm interested, and we are interested in knowing about, you know, that part of the path, because in a way, you've done what many suggest professionals in the industry do, which is, hey, you need to adapt to AI. You need to find what are the good things that you can do with it, et cetera, et cetera. So when you. You first transition, how was that meaning? Did you have, or did you share the common fears that people have today on. On artificial intelligence? How did you go about that process?
[00:14:07] Speaker C: I wouldn't say that the path I took is the common path, because both me and Belen, we are, in a way, pioneers. We were trying to figure things out, and there was an element of fear because back then it was still not known how powerful AI is, and we still didn't know what exactly it is good at and what it is bad, bad at and how things are going to develop. So my path was really pure exploration, without any expectations and without the fear elements. On the contrary, I think I was, and still am extremely excited and inspired and positive about it, simply because it's so very interesting.
So I wouldn't say that this is the path that most linguists are going to take, because in a nutshell, our path is a path of exploration, whereas most linguists and translators will have to take the path of learning and adaptation. And that is a different perspective. That is a different way of entering the AI journey, adapting to processes that have already been developed or technologies that have already been introduced in workflows and accepting them and adapting to work with them. That is going to be the common path for the freelancers and translators and all professionals that are involved in the localization industry as part of the production, not the research.
[00:15:53] Speaker B: Yeah, I totally agree with that, Marina. And we are somehow in a privileged position also because we work really hard, hard to get here. It's not just for the sake of it. But I agree with Marina that we are in a privileged position because, yeah, we were there at the very early steps of how the technology was being developed. It was more about curiosity, learning. So it was more about, yeah, learning something New trying how to influence it for the best, and so on and so forth. So I shared that with her, but it's true. And also, Marina is totally right. When we started working on this, before the gen AI explosion with ChatGPT and so on and so forth, the hype was not so high. So high, right? Like the hype was okay. There was some conversations, but it was still kind of okay. We are, as Marina said, we are exploring, we are seeing what this can do, and so on and so forth. But now the hype is so, so, so high. It's so high that it's going down already, according to Gardner as well. And I totally agree. I see that as well when we go to conferences that people are now more realistic and adjusting the expectations and starting to discuss what we can actually do versus what we cannot do. So that's good. But there's still some fears that I have, some concerns that I have regarding the evolution of our industry and also ethical concerns that are shared by freelancers as well.
Because there is as much we can control on the industry in the sense that if the message that AI can do everything as good as human, and maybe it will at some point, but not today, if that message that everything can be automated, that you don't need humans, as we've been hearing some CEOs from some companies saying that in three years we won't need human translators, if that message starts really permeating the society and everyone interacting with translation, that's a concern to me because it's a little bit more unpredictable, how we will evolve. There are many people working in this industry as translators, thousands of them. So telling them that in three years you won't be needed, it's a bit harsh in my opinion. And I think we need to be a little bit more empathetic when we talk about it, because fears are real and we are human. We are still human. We are not AI. So we have feelings. And I think we need to manage this change and this transition with a little bit more empathy as an industry as a whole. And that is a fear for me, that we will destroy something that is beautiful today because not knowing how to address these changes, how to get people on board, how to make it exciting. But anyways, luckily there are people like Marina, for example, who are very optimistic and sharing really nice stakes on how the future could look like. And I think that's what we need. More and more people who envision a future that is actually sustainable and that is actually exciting for everyone.
[00:19:28] Speaker C: And Belen, that's actually the only fear I have, because you started by saying that we were in a privileged position because we started early with our AI exploration and we were spared the fear this brings. This privileged position to me has now transformed into a position with enormous responsibility.
So I don't feel that privileged anymore. I feel extremely responsible to take the right path and responsible for preparing translators, freelancers, the community, the industry for what is to come. And that's how you end it. Your response, like, we need to know where we are going, make the right decisions. And the fear is that we will not succeed in that, that we will not predict correctly, that we will not prepare translators, linguists, and everybody involved for what is to come. It is our responsibility to educate, and it's our responsibility to share the learning. We cannot expect that everybody will take the path of Belen and mine and spend weekends reading articles and investing their own time. It is on us to actually help the community grow and be prepared for the new jobs to come, for the new tasks to come for AI and.
[00:20:55] Speaker A: Thank you both for sharing that in our experience. Some of the. Of course, the biggest fears are around what we had talked of, Mike. Right. Job loss and in some cases, deskilling.
But from what I hear from you and what I've heard from other conversations, it feels more like an upskilling opportunity. And then the biggest challenge there, or the biggest threat there for the ecosystem is that many people disregard the opportunity of being the translators of cultures. And then we talk a lot about culturalization. So if that kind of, like, layer of knowledge disappears, meaning translators and interpreters that are very well versed in the nuances and context of when translations and interpretations happen and why they happen, et cetera, et cetera.
If we lose that layer, that means we will lose a lot of time. We will waste a lot of time just because we don't know how to manage that transition.
Blenn Marina. Do you think we can do it? Meaning, okay, let's say there's a lot of people with responsibility, a lot of people talking about it. We are talking about it. People are proposing ideas. Do you think we can do it, or are these fears very justified? There's still a long time for us to get to where we should be in terms of the role translators and interpreters will play in the training data, as an example, or like the ecosystems that facilitate the training data that make the artificial intelligence help us do what we do.
[00:22:39] Speaker C: Okay, there was a lot in that question, so let me try to kind of decompose that and start simple because it's a very complex question. So I will simplify the task by decomposing the work of a translator. I think a common misconception is that we think of it as a monolithic piece of work, right? So you translate, that's it. It's like one block. You do translation, but translation is not one block. It's not like one activity, it's a process and it contains tasks. So it is decomposed into different tasks. Now we look at each one of those tasks. This is the right way to look at the translation process. It contains blocks of individual tasks. And how do we handle one of those tasks? You call them layers, I call them tasks. For each one of those tasks, we can have 100% human effort or 100% technology. It can be AI powered technology, or it can be something like, as dumb as, like, I don't know, spell check. I mean, it's not that dumb, but you don't need a lot of AI to do spell check and you don't need that much human intelligence, right? So if you think of the translation process as a sequence of individual tasks, you can think of implementing technology on some of those tasks only. What is changing right now is that more and more of the tasks in that process can be done automatically. Whereas before just spell check was a helpful translator. So this is just one of the tasks a translator did. Now it's spell check and terminology suggestion. And you know, with AI you can have style guides, enhancements and maybe automated review and so on. But still, even in that process, there will be some tasks, components that are going to be 100% human because we have the human layer that is irreplaceable, that is the genuine emotion, that is adapting to culture specifics, that is bringing in the rhythm and poetry in a translation. You know, if you're doing some translations that are, let's say marketing or high end translations addressed to audience that needs to be enticed or in some way attracted, you know, even to click, sell, buy, button, whatever. You will have the human 100% involved in some of those very complex tasks where AI is not getting anytime soon. Okay, at this moment, LLMs are really good at language, but they're missing a lot of stuff, even from language. No, but they're missing a lot of stuff when it comes to theory of mind, psychology, intuition and that sort of things. So to me, the evolution of the translator role will go two directions. One is focus more and more on those components from the translation process that are human centric only and can be Delivered that work can be delivered by human only. And the second branch would be become proficient working with AI so that you can guide the AI to do the right things when it comes to those components of the translation process where AI can be applied.
In other words, translators will keep some of the work they're doing, some of the components of the work they're doing today because they're irreplaceable there. But they will add another type of work. That's the upskilling part where they will need to guide the AI the right way so that for the components of the work that are to be done with AI and technologies, translators get the right results.
[00:26:33] Speaker B: Yeah. Fascinating. Marina, always listening to you and how you super logically break down everything so that it makes sense. I want to take a different take on that. So I feel right now both LSPs and translators are in a similar boat. And I want to challenge Renato. Renato, if you're listening to this, please tell me what you think. But you know, I'm a very big fan of Renato's and Tucker's book, the General Theory of the Translation Company.
And what Renato has been saying for a long time is that, you know, LSPs, they don't sell translations, they sell project management, vendor management, what is it, sales. And I don't know everything by heart, but you have to read the book, it's a really great one. I totally recommend it. But I have the feeling that somehow this might be changing because of AI, because we are seeing how more and more and we've been having conversations about this, Marina, and with other industry friends, but how developers and tech people, IT departments within the localization buyers are trying to solve the language problem, right? Trying to solve the language barrier when it comes to localization. And the thinking goes, okay, we have this technology, we have this gen AI and all these things that IT can do so we can automate everything. We don't even need a localization department anymore because we can do this automatically and so on and so forth. So I'm not saying of course that is true, we know it's not. But this is where some of the reasoning and thinking is going in the industry. So I wonder, even for LSPs and I work at Terra right now, more mid sized lsp. So what I would also suggest to translators and to all mid size small lsps to make this exercise, thinking more about the future and okay, what is our value proposition today? What is what we do today? And if the technology gets so great at some Point what it is that we add to the table, which are the adjacent services, which are the niche, really super hyper niche places where we can keep adding value. Because even if we, as Marina was saying, okay, we focus more on the AI, sorry, on the more human centric layers or tasks within the process that will also be very, let's say down in the supply chain. So it will be more on not always. And Marina, you can reply to me later. But more on the validation side of things, more on the validation. And I think that will only make that translation services are more and more commoditized. They are super commoditized already, but this new wave of technology and hype will make them more commoditized. So where can we actually add value? I feel like the future is not so much on the translation scale skills per se, because translation skills will be more like mastered by the technology, but more about the subject matter expertise, which is basically similar to what Marina was saying. So that poetry, translation, that creative, that thing about conveying emotions in a marketing campaign or user experience, being experts in psychology, in communication, in marketing, in the actual fields where we are working, rather than just on the language itself and how we can support brands really connecting with their users, buyers or whatever it is, audience, readers, players, how can we support brands there being this hyper niche subject matter expertise for certain verticals, for certain cultures and so on and so forth. So I think we will need to evolve to, yeah, experts not only in language but in going upstream basically for our customers. So what are the conversations that are happening upstream and not just enough afterthought in localization as we are living these days.
[00:31:16] Speaker C: Okay, Belen, I have to take over because you said so many things I want to respond to. And I will start with a vision I had this morning while biking to work. Okay. When biking to work, I always listen to a podcast or a meeting recording. But this time it was a podcast between a physicist and some AI guru. Don't remember the name, but two, I don't know, former Google AI, whatever, very important people. And one statement that caught my attention was talking about how the work of software developers has changed over the years. So when those two guys were young, apparently the way they were writing computer code, software code, was writing it, like really writing it. So you would sit down and you will just write all the commands and so on. The way software programming works today is nobody really writes code. People assemble code. They take ready made blocks and they assemble code and then they polish it and they add this additional polish. The final touches to make sure that it works really the way it works. But nobody writes code from scratch. When I heard that, it dawned on me that the translator work is probably going to evolve in a similar way. Before, translators were writing translations, right? Even, you know, with pen on paper. Okay. And then, you know, typing it in. Then they got tools that help them be faster by post editing already pre written translations. Okay. And then I thought maybe you'll think it's a crazy thought, but what about having the translator's work evolve into assembling translation? So imagine having a sequence of AI prompts that generate different components of a translation. And you assemble. You assemble, you for example, have the basic translation and then you add to it the prompt for getting the style right. And then you get at, you know, another layer of getting the terminology right. And, and in the end you add this human touch to get, you know, the feelings right. So you basically build it up the way a computer, a software programmer assembles code, you assemble the translation. For that you need a lot of subject matter expertise and a lot of human knowled. Right. But the work changes. That was my vision. And I really think that this will bring the translation profession to a whole new level when it comes to efficiency, but also when it comes to the creative process. Because I think that not wasting time in just typing out things, but having more possibility to compose things and assemble the final language product in the end, it's a more rewarding type of work. What do you think about that? Does it sound too crazy to you?
[00:34:22] Speaker B: No, absolutely no. Marina, I love that idea. I love the analogy with coding. And yeah, I think this is kind of combining the two ideas that we were sharing somehow. So yes, why not?
Yeah, you made me think about it. It could work and subject matter expertise to assemble translation, but also to depending on the content, because we're talking in general. But then each content type, each use case is different. But I'm thinking more about video games, which is what I work more on these days, and really understand one of our customers, they have such a specific jargon and a specific way of thinking, so super hyper hyper niche that a human understanding that language won't know what they are saying. Because if you don't know that so deeply, you are not able to communicate what they are trying to say. So I think a combination of, yes, assembling not only the different layers of the language aspect, but also the holistic experience of each content type could be something very rewarding. And to me, I'm envisioning a future where the scope of Translation is wider and not more narrow to just clean the AI mess, which is things that are being said. It's more about what is the holistic experience and how can your niche subject matter expertise add to that layer because the basic thing will be done by the machine. So I love that vision of this assembly and future. Thank you for sharing, Marina.
[00:36:11] Speaker A: Thank you both for engaging in that awesome conversation. And we are already touching upon what are some of the things that you see people doing in the future. Right. We start seeing or we start talking about some of the evolution of the roles, to put it that way, and the sophistication of some of the functions that already exist there. Along those thoughts that you have about how those roles are going to look like in the future. If you want to delve deeper into that, please do so.
What is the evolution of those roles that we have today and what are the skills that are going to be most valuable in this adaptation? And I fear to say just AI because I think it's an adaptation of so many other things that you were putting out there, Belen, in terms of like our interactions with, with end users and even with, with other clients along the way. So any of you, if you could, if you could help me out with this with these thoughts.
[00:37:12] Speaker C: Oh, that's the toughest question to answer. Like, how will the translator job evolve in the future? Because there's a lot of guessing, right?
My intuition is, my best guess is that first of all, there will be a certain percentage of translators who are translating very special content who will continue doing it more or less the same way, because there we need the human touch, the human validation, the quality.
And it's going to probably be like 5% of translators. They will keep doing what I call artisan translations, like really boutique translations, like, not the mass industrial translations where we churn like millions of words per day. Belen the other day actually tried to put some numbers on that distribution. You know, how many translators, like what percentage of translators will keep doing this purely 100% human work.
And then I think the mass translations there we will see more and more technology.
And that's where the role and the responsibilities of translators will shift. I actually wouldn't even call them translators anymore. They will need to be a combination of, first of all, linguists. And by linguists, I mean somebody who deeply understands the language at almost a mechanical level, because LLMs, they have a mechanical level understanding of language.
Prompt engineers, even more so. Like they try to decompose language into mechanical rules, more or less, but then already mentioned that.
And we need linguists to be able to talk to them, to communicate with them and also have a representation, like mental representation of language on a more kind of theoretical, mechanical, like syntactical, morphological level.
So that's one skill they need. And the other skill that they will need is tech savviness. I think it's going to be inevitable that linguists are more aware of the more exact way of looking at data, linguistic data, language in general, and also have some understanding of computer code.
Not produce it, but understand when reading it. So I think that the necessary skill is going to be technical savviness, data analysis. That is super important when it comes to AI. The ability to orient oneself in a data set and kind of learn to operate and transform data. And those two skills, those two new skills, let's say, even though many translators have them already, will allow them to basically collaborate with AI in a new type of localization processes and translation processes to get the maximum out of it and still keep the fun in the work, right? Not just become machine inspectors, you know, trying to find an error, but actually actively work alongside AI and with AI to produce language content in their language.
[00:40:44] Speaker B: Yes, Marina, fascinating. Thank you for sharing. I will add something beyond the purely technical or linguistic skills and I think something that is there is a very big gap in the industry right now, especially for freelancers, but also sometimes even for, for some LSPs, is the business knowledge. Guys like you need to be aware of how your business works, of how technology is impacting your business or how things are evolving, macro trends, micro trends. You know, you have all these great sources like multilingual Nimzi and others that are doing research, posting their thought leaders on LinkedIn and so on and so forth. So really understanding that if you just stay there waiting for the industry to evolve in a way that you like or that works for you, I don't think that is going to happen. I also think that you need to analyze what you want to do with your life. And if you really want to work with language and just do it and translate from scratch, then go and find the customers that want that service. Maybe lsps won't be that customer anymore. So you need to be more proactive finding those niches. Again, I think hyper niche is key here and there are many great translators out there succeeding, having a very good living because they are more proactive in designing what they do. I know this is difficult. I know this is. Sometimes you just don't have the resources, the money or the support to do this, but really you need to start thinking about what you want to do with your career, how you want to shape it, what type of work you can do, what type of customers. Sometimes it's better to work with a local company in your city or in your country than working for a multinational in the United States just because it's a cool name. But then you're not aligned with the values. This is something that I've experienced myself this year, and sorry for losing the focus, but sometimes you need to find what your values are, understand what you want and go for it. So the industry won't tell you exactly which steps to follow, but you need to really understand.
Yes. What you can do with what you have with the circumstances of the industry and find that way and diversify your service offering. Maybe you don't want to be a post editor or you don't want to be.
To be involved in these technical aspects of localization. Well, find something else. Reskill, upskill, learn something else and keep adding value. Because I think the knowledge that a translator has is so important for society in any aspect that it cannot just be lost out of losing faith. Because we don't like the industry because things are changing in a way that we don't like. So that is something I want to say.
And the other thing, apart from what Marina said, is really finding those niches. Like the other day we were in Brussels and we were talking with Adrian Probst from Freelanceverse. He has this great YouTube channel. And thank you to him and people like him who are sharing what they do and how they succeed, because that's. Some freelancers are alone at home and don't know exactly how to go about things. So kudos to Adrian. But he was explaining to me that he's very niche because he works into Swiss, German and in a very specific vertical, which is sports. And he found a way to add value there. And it's not something that happens overnight and it will probably change over time as well. So it's not like you find your niche and forever it will work for you because we live in such. Such a changing and evolving world. But I think really finding those places where you can add value when you can actually say, I'm better than the machine. And this is the proof of it. And that's something that we all need to do. And including LSPs, how we can stay relevant in the age of AI and keep doing what we love. Basically.
[00:45:27] Speaker A: Yeah. And that's a challenge for everyone. It feels like one of the expressions that I use that. I prefer not to use it often, but it's business as usual. And sometimes when you're in business, clients don't want to renew. And it's almost like, what do you say? Then you have to adapt and you have to do different things. And it feels like it's probably a good situation to start taking control over some other conversations which we will have in the future. But, you know, we have this area of our conversation that I hope we get some time to talk about it right now, which is the future of talent, the future of work, and of course, where we get that knowledge to be prepared. And you have an academic background. And then sometimes when I hear like, oh, that, you know, the next generation, I feel bad about myself because I'm also constantly learning. So I'm also thinking myself, okay, like, what are the things that I should be looking into? What are the specializations that I could be looking into that can help me do my work better, right, in communications and journalism and whatnot. But what do you think? And you've been in the academia, you've learned about linguistics, about translation.
What. What type of programs, what type of different areas of knowledge need to be explored further in academia to prepare talent for what's to come? From your perspective, of course.
[00:46:58] Speaker C: Thank you for asking that question. It's a very, very important question. And me as a. As an educator by heart and a former academic and teacher at universities for master students, I really have very strong feelings about that. So I think that the curricula of translation studies at all universities where it's offered must be updated immediately as of next term to include a course on AI. What are LLMs, what they can do so that we set up a realistic expectation among students and prompt design how to talk to AI. There are specific ways in which you need to communicate with artificial intelligence varies a little bit depending on which exact model you communicate with. But there is a way to talk to artificial intelligence different than the way you talk to a human. And this is a skill which all humans who want to be involved with the new technologies must acquire. Students of translation studies most of all because these are large language models. Translation works with language, and it is an absolutely mandatory skill.
The other mandatory skill, in my opinion, is some knowledge of data. Data analytics, bi.
How to operate SQL databases. Language is currently seen as data, okay? And it will continue to be data as long as we have AI involved, because that is the basics for training models. And I think translators must understand what data actually is, how to work with data, because that's going to become very important in the future.
It's going to be a necessary skill when, for example, being involved in helping train models, fine tune them, let's say creates validates data sets or creates new data sets to be used for training language models on edge cases like, you know, the long, the long tail problem. All of that will necessitate linguists, currently translators, as I said, I don't think they'll be called translators anymore, but it will be the same people who have the, this will be the people who have the same skills that translators have today.
They will need to know how to work with data. So AI prompting mandatory university course, I think for any student, no matter the discipline. Right, but for translation studies, for sure. And knowledge of how to operate with large data sets, how to transform them, how to manage them and how to read them is the other mandatory skill.
Totally.
[00:50:00] Speaker B: Marina, thank you for sharing. I mean, I agree with you that curricula at universities need to be updated as soon as possible. Actually, Marina, the other day we were discussing this European funded project called LT Leader Project where a consortium of different European universities are going to work together to see how, you know, the curriculum needs to be updated, what's the future of the profession, and so on and so forth so they can collaborate in basically revamping the curriculum. At the same time, I want to say something because I'm very European when it comes to university and I cannot avoid to say that, you know, universities, the. They are kind of the last bastion of knowledge, right? And they should also be somehow free of teaching and free of teaching things that might not be super productive in the world these days. And not everything needs to be productive, but sometimes knowledge for the sake of knowledge is still needed. And I think we don't need to lose that somehow. Of course people go to college to then be able to have a job, right? So you want that to be useful. But sometimes I think we put too much pressure on universities and academia when they are doing what they can. They have so much bureaucracy.
And again, it's not the role in society to just create productivity machines, human machines, but just to, to protect freedom of thought and knowledge. But I know that's a very European and idealistic way of looking at it.
The reality is that in the United States many language programs are closing in many universities because no one wants to study languages is also happening in the UK and unfortunately I'm hearing from colleagues from European universities that, that more and more translation programs are at risk because universities also want to monetize and so they want to make money, so they want to create programs that appeal students. So, yeah, I think that's a challenge that is currently happening in academia. And I think for example, in what Marina was saying, including more AI related topics and making language programs or translation programs more holistic and widening the scope of these programs can be a way to protect them as well.
So, yeah, it's not an easy question again, but I definitely think that technology needs to be included there and students need to be told that technology is going to be a fundamental part of what they are going to do and if they don't like it, maybe they need to study something else.
[00:53:12] Speaker C: I want to say, Belen, thank you for bringing in your European perspective on education and the role of universities. I cannot agree more than possible with you because I'm European as well. I just wanted to mention that what you mentioned, that universities should be also free to research without having an immediate utilization and kind of profit from that. That is very true. This is the importance of ground research. And you as a former researcher and me, we know how important it is to be have the possibility to research things for the sake of simply researching it. And you never know when that research will turn out to be useful. Maybe in 20 years from now somebody will read that article and that will be the right time to apply the knowledge that you discover today. But you cannot apply to anything because the technology is not there yet. So ground research, definitely. Which actually made me think about where that ground research must happen, whether in the linguistic departments or some other departments. Because currently the ground research that we need is simply explore the interaction between language and large language models. This is what is relevant for us, right? Like just, you know, explore things such as, you know, those niche questions. Does it pay off to be polite when talking to an LLM, are you going to generate a better response to your prompt if you say please or not? Like, these are really, you know, who really cares about that, right? But it's important to research that. And one thing that I found discovered recently, and I thought that was very interesting, is quite a lot of AI research which is done at the University of Prague. And the most interesting articles that I read were actually on AI and language and they were coming from the math and physics departments, not from a linguistic department. The math physics department at the University of Prague has a kind of a, I don't know, division that is doing natural language processing, but it's under math and physics. And they have extraordinary interesting research and findings and very, very interesting people over there. So maybe I have a soft spot for math and physics generally because of my background, but I thought that was a very interesting thing.
[00:55:34] Speaker A: Yes, you definitely have a soft spot for math and physics, that that is for sure. And I know we're coming to an end of our conversation and we should definitely do this again, but I want to be very respectful with your time and I'm very thankful that we have been able to talk today. Are there any final thoughts, comments, messages that you want to send to the industry, to the ecosystem, about localization, or anything at all that you want to mention that we didn't have an opportunity to talk about?
[00:56:04] Speaker C: From my side, it will be just an appeal to the community of linguists out there to be open minded, to give AI a chance to go out and learn, adopt the new technologies or at least explore them. And they can start by exploring AI in their daily work. I mean just, you don't need to do it as part of your job. Just try it out, see how it reacts. Because by doing that they will learn that there's a lot of potential, a lot of things that AI gets right. But there are also a lot of things that AI doesn't get right. And the interesting thing is how to help that AI gets everything right. This will be the role of the linguist, like help technologies evolve and produce better and better results so that they are really helpful. That would be my appeal to the community. Go out, explore and learn.
[00:57:04] Speaker B: On my side, I would like to say that let's not let technology and all this polarize our industry because that's what I'm seeing these days, a lot of polarization, a lot of black and white, even in society in general, right.
A lot of you are either pro AI or against AI or either is horrible or really good. So move away from that polarization, have meaningful conversations, Conversations as an industry. Also speak up the truth, be transparent and be open to have the difficult conversations. Because everything has consequences. We are all part of this ecosystem. We are really passionate about our industry and what we do. And everyone, I think wants to really make something beautiful out of the localization industry. So. So yeah, let's have conversations. If you are worried about something before letting fear or frustration take over research, talk to people doesn't need to be conversations on social media in front of everyone.
I invite anyone who wants to have a chat to reach out to me, to just have conversations because this is, I think what will let us evolve and manage this change and this transition in the best way. If we keep being human and communicating and being empathetic on what we do.
[00:58:44] Speaker A: Thank you, Marina. Thank you, Belen, for joining us today. This was our conversation with Belen Ari from Terra Localizations. Marina Pancheva from rws. And my name is Eddie Arrieta. I'm the CEO here of Multilingual Magazine. This was localization today. Marina, thank you so much for being here with us today.
[00:59:05] Speaker C: Thank you so much for having me. It was a pleasure.
[00:59:07] Speaker A: Belen, thank you so much also for joining us.
[00:59:11] Speaker B: Muchas gracias. Such a great pleasure to be here with you and Marina.
[00:59:18] Speaker A: Excellent. And everyone who has been listening to this conversation, thank you so much for listening once again. Until our next time, goodbye.
