Episode Transcript
[00:00:03] Speaker A: Hello, welcome to Localization Today where we explore how language, technology and community converge to unlock ideas for everyone everywhere. I'm Eddie Arrieta, CEO at Multilingual Media, and today we are talking about the role of the human in localization workflows, what augmented humans who really means, and how companies are navigating the balance between automation and human creativity. Our guest is Vasilis Hamalidis, CEO of Alpha crc, a company known for shaping localization services through deep tech human collaboration. Vasilis, welcome and thank you for being here.
[00:00:52] Speaker B: Thank you for inviting me, Vasilis.
[00:00:54] Speaker A: I already said it right before we were talking. I'm very excited about this conversation.
It means a lot to the localization industry, to language professionals, to linguists, to globalization professionals, to global content creators. It's a conversation about the human voice and we're very excited to have it.
So you've built a career in spanning global corporations and localization companies. How has these diverse backgrounds shaped the way you lead Alpha CRC today? Welcome again.
[00:01:28] Speaker B: Yes, thank you again.
Well, I actually grew up in a household where my parents were translators and interpreters.
I was born to Greek parents in Germany, so I grew up with German at school and Greek at home. And so translating Greek became sort of like a second nature.
And while I didn't pursue that as a career, the thought of communicating in various languages has always been with me. And so I've always enjoyed speaking different languages, not only to speak them, but to communicate with people. And I think what we are talking about is really that how can companies communicate to their audience in the most appropriate way, independent of where the customers are? And can you do this with technology? Can you do this with people? Can you combine both? And what does it take really to be efficient and effective in the process? I think that's really the big question that we are facing going forward. Where is this taking all of us? Where is it taking the translators, these beautiful people who invested so much time studying languages and techniques to translate, often not helped by their universities because of the lack of up to date technology, but also not helped by the customers, by the LSPs, by their environment?
And how can we make sure that what we are producing is the most appropriate content for our clients? And how can we provide the right support and the right tools to the linguists to help us generate the right content?
[00:03:28] Speaker A: And that's a really great conversation because that content is transforming, the volumes are changing, the cues that you're getting from it is changing. And as you were presenting feels very human. It feels like human evolution taking place right in front of our eyes and having the opportunity to observe it, but also document it and at the same time be part of it, I think is very exciting. There's a lot of talk in the industry, localization, translation, globalization, all of it, of augmented humans, what you're mentioning, right, like that sort of evolution. What does this mean in practice?
[00:04:09] Speaker B: Well, let's say that to me, an augmented human is a bit like putting a vest on, on a human that's equipped with all kinds of gimmicks and tools that the human can push or like a binomic hand, and that catapults the human into an ecosystem of intelligence and intelligence tools and databases. And the human can use all of these things to get things done in a much quicker, more consistent and efficient way, basically achieving higher output, better quality in a much tighter timeframe. I think augmenting means increasing the human's ability to perform.
And that's something that we acknowledged a few years back. We actually developed a, an internal tool to help our linguists, we call it Omnitool, which basically provides the linguists with AI capabilities so that they can generate knowledge much more quickly when they have to do research, that they can use AI output to help them go faster to check things. And, and once that's done, the result is really something that's faster, better, more economical, cheaper to sell to the client, and a great support for the linguists.
When you think that we used, I mean, Alpha was set up in 1987 as a in house operation. Isabel Weiss, who is the founder of Alpha and was still our linguistic director, she had this vision to have many functions under a single roof and have people working alongside each other. Engineers, ATP experts, linguists, project managers, et cetera. Just five years ago we employed 200 linguists and today we employ 120 linguists. I mean, that gives you an idea of where things are going and the volumes that we deal with haven't really changed a lot. So you can deduct from this that the productivity of the linguist has gone up by, you know, almost, it's almost double from what it used to be.
And that shows you how important technology has become.
And augmenting the human means really helping them do things more quickly and better.
I would say, however, that it's not something that all of the linguists can benefit from.
We are being faced with a number of people who can't absorb technology and they're being left behind and ultimately they have to look for an alternative path in their lives because they can't really get augmented if you wish. It's not something that everyone is made for to work with technology and to accept technology.
And also some people feel like there are artisans, there are artisans who craft words and content.
And once you introduce technology, they might feel that this is not what they were educated for or that's not what they like to do. And so many of them have been walking away from the industry, some have walked away because they can't find enough work, or it's not work that they like to do, not the type of technology they like to use. So all I'm saying is you can augment the human, but not all humans can be augmented.
[00:08:11] Speaker A: That's a really good way to put it. And of course, there are tasks that even they might seem as trivial as to responding to emails, but it's the big bulk of. The most recent ChatGPT usage report by the National Bureau of Economics Research was talking specifically about that. And it's. And it's how. Just chatgpt. We're not even talking about any other artificial, like fancy things. We're talking about ChatGPT. It's like there is so much work getting done there. There's a lot of work, a lot of emails, a lot of summaries, a lot of. A lot of work is getting done there. And there is real economic value and you just validate it as well. You, you have teams that are twice as productive. That is incredible. And there is this pressure to automation, right?
[00:09:03] Speaker B: There's absolutely pressure, but it doesn't stop at the linguistic level. Just think of project management or account management, people who have to interact with clients. And we had cases before where some of the project mentors were not really fluent in English and someone had to review their emails before they went out to a client. Now they use AI and they write beautiful emails. I mean, granted, they might not be as creative as some other writers, but in terms of progress, it's tremendous progress throughout the workflows and throughout the value chain. It's not just on the linguistics side, it's when you think about the. Let's take recruitment before, a human had to review the CVs or resumes and decide what to do with them. And a lot of time there was either bias or there was a lack of consistency in the review. All of a sudden you get the possibility to have a much cleaner review of things with the help of the machine. And that allows you to go faster in the selection process and invite the people you really want to interview. So I can say with conviction that it is very helpful. And we can't talk the technology away. It's here and it's not here to disappear, that's for sure.
[00:10:30] Speaker A: Absolutely. Absolutely. It's not going to disappear. There's no way. There's too much productivity that is stemming out of it, allowing many human beings, especially those that work in the digital realm, let's put it that way. But it has ripple effects outside of that digital realm.
And I can also talk with conviction. I'm a part restaurant owner, and we use artificial intelligence to analyze our market.
So we feed it what our competitors have, the breads that they sell, their marketing campaigns, all of it. And we say, this is what we see. Is there anything we're missing?
What do you see? A perplexity. What do you see in other countries that others are doing? So the learning, the way in which we learn, it's changing significantly because of the access to more synthesized information.
[00:11:23] Speaker B: Exactly. And you think that some professions are shielded from this, like you would say a plumber or electrician. Why do they need AI? I had a carpenter visiting recently, and there was a problem with a deck. And he pulled up the deck and looked at it and concluded that it had something to do with the humidity levels. And then his son, who uses AI pulled up his phone and asked AI the question why it was the way it was. And the answer that came through was exactly the right answer. And. And so the father was totally blown away by that. And it shows also the generational gap that exists in adopting technology and understanding how to use it. And it's just an example how even the most manual professions can be augmented if you wish.
Experiences there is very valuable. But then, of course, the question is, what will happen once the experienced people leave this world or the. The job market? Who's gonna have that experience and expertise in the future? And that's. That can be a very scary thought as well.
[00:12:36] Speaker A: Yeah, yeah. Cause it's not. And I was thinking about it as you were speaking, when we were talking about translation. It's like, do you really understand that translation? And let's say there is something translated from Greek into English. The way you read it, you know, in Greek, and me not knowing any Greek, am I understanding that translation? And I'm reading it in English. We could find many examples of.
And it will be like, I will not be able to really understand that translation, even though I can literally read the words that are there. So there is that layer that cannot be lost. I believe if that's lost, then we're Losing an understanding into a human behavior itself.
[00:13:17] Speaker B: But I think we are seeing some evidence that translation has always been bought more as a commodity than it's something very valuable. And it's one of the very few services that require using people with a very high degree of education and being paid like a commodity. And it's really not fair to the people or to the companies providing the service. But it's a fact the arrival of AI has only worsened the situation because now people, the clients, for most of the clients, they are not in a position to judge how good a translation is. And they might have some people in country, but some clients don't.
And so now they think that because there is AI. Well, why do they need to pay for a professional translation service? Why can't they just use the LLM to produce the translation? And so, so it's up to us really to make a case for involving the humans in the production process and what kind of benefit that represents to the client and is it worth the investment.
I would even argue that clients should segment their content and say, okay, this is low impact content. This can be done by the machine, but this is high impact content. The machine should only be used as a support, not as the main driver for it. And the other thing is, I mean, has anyone really thought about the cost of using AI? It's not really free, is it? You know, once you take into account the tokens and the electricity used and whatever other fees are charged by the providers of AI, not to mention the environmental impact, it doesn't come for free, not really.
So whatever you do, there is a cost associated with that. And if you use AI in its raw form or machine translation, which is nothing different from really basic AI, it does come with a cost tag. Despite being marketed as free sometimes, and.
[00:15:38] Speaker A: Thank you for putting it that way, makes me really think about the augmentation that you're talking about, like going, going from that. And as you were speaking, I was thinking, yeah, that translation as a commodity and the cultural nuance was assumed. It's like, oh, you're going to have cultural nuance. And it's like, okay, once you go to the machine, it's like you can assume the cultural nuance. The machine will have to be given the cultural nuance. It would have to be trained in so many ways to be able to do that.
[00:16:08] Speaker B: Exactly. And there's a really significant contribution by the linguists that's possible here.
How they prep the machine for the output, what kind of prompts they provide to the machine.
This is where the difference happens. And so the output from a properly prompted and trained machine versus this generic LLM output can be very, very significant.
Let's not forget that a generic LLM has been fed with so much data from everywhere that it doesn't mean that it's relevant for the very specific requirements of one single corporation.
So if you are the client and you want something very specific to you, well, you better make sure that whatever you use understands what those requirements are, what your linguistic assets are, what your tone of voice is, what you know, your brand, your feel. And if you know, if the machine can understand that, great. But if you assume that a generic machine will provide you with that, I think you're mistaken. And here's where the linguists can make a difference, where the LSPs can make a difference.
Whether, you know, in the future these companies will call, will be called lsps or something different.
It's a different story. But certainly we can contribute to making the output more specific, more customized for the client and getting better returns for the client.
[00:17:53] Speaker A: Absolutely. And I'm really glad Vasilis were having this conversation. I'm also sure our audience appreciates the conversation.
So there is that pressure and I mentioned it earlier from the automation side, there are things that will be definitely automated and the pro will need to have different approaches from different angles. How do you see that conversation, Vasilis, regarding automation, of course, with the use of artificial intelligence and what would be, what would stay in the future as augmented or is it augmentation, a temporary thing and then again will disappear?
[00:18:34] Speaker B: Well, I mean you can augment and augment and augment. You know, I don't think the definition of augmented today is the definition of augmented tomorrow.
We will reach some point and then we'll say, okay, what's next? What now and how can we go further? And then we will start again to questions what we are doing.
So the augmentation today is all about how can we leverage technology to help the linguists produce better and faster.
We have developed a number of tools in that regard and with those goals, the range from having quality evaluation systems that allow you to send pre translated segments in two directions. If the system thinks that the segment is well translated, it goes straight to publication. If it thinks that it needs review, it goes to review. So you can save a number of, well, a number of hours by the linguists who don't need to look at the already well translated segments any longer and they can focus on the ones that need to be reviewed. As an example, we've developed multilingual term extraction, which again provides a much faster way of getting to the right term base than if you had to do this manually.
And we are also talking about using private AI models to preserve the confidentiality and the privacy of the client's data, which is still one of the reasons why you see human translation, or at least post editing, being in demand. If we didn't have those reasons of customizing the translation for the client, of needing a level of trust to work with the client, you would have seen human translation being almost extinguished by now. But clients need that sort of trust. They need to know that you're doing the right thing for them. And they need to know that you take their requirements seriously, that you understand their pain points, and that you offer some technology solutions which are not even products, they are real solutions where you take into account you're not selling a product, you're selling something much more integrated than the product that's composed of some technology of a human and combining those things and being relevant to the client's pain points. And if you can offer that to the client, then I think that human involvement in translation will be there for a period of time. Whether it will be forever, who knows. I don't think we can predict that yet.
But it certainly explains why the disappearance of human translation has been much slower than some predicted even two years ago, and that hasn't materialized. We also see that some clients choose cheaper methods of translation production, but they pour more content into localization, content that previously remained in the source language, whether that was English or Chinese or whatever. We also see much more content coming from other source language than languages, than English.
I just recently looked at the statistics from before COVID where we had a total of 103 language combinations, and all of a sudden we have 224. Why is that? Because the number of source languages has increased and the volumes from those languages as well. All of these things explain why the LSPs are still around, even though I'm not going to deny that things have become much more challenging. And if you don't adapt, then you see a lot of the smaller players disappear, being bought up and having generally trouble figuring out what to do in the future.
[00:23:01] Speaker A: Yeah, that's, that's, that's really telling, I have to say. There is then still demand for human translation. And I think I can infer from your answers, and you've even talked a little bit about this. What do you think? Human translators and linguists, perhaps people. We should have asked you initially what is A translator to you, right? A person that goes worldwide. I don't know what it is, but what have the translators, what do they have going for them? Which I think it's something that a lot of them would want to hear.
[00:23:33] Speaker B: Yeah. When I look around within Alpha and then when I look at our freelance partners that we have, there are people who stand out because they actually, it's a mixture of expertise, sometimes subject expertise.
There's certainly areas of translation that are still shielded from a large degree of automation. Let's say in healthcare, where medical trials or pharmaceutical companies for patent translation still rely heavily on humans because they don't want to take any risk because they think it's less risky. Whether that's true or not true, I'm not judging that. It's just the way it is. So if you have a particular subject expertise in an area that's particularly in demand for human translation, that's great. But if you take for example, the legal profession, AI is very strong in the legal field. So rather than doing translation, you're being asked to review and post edit things. And this is where things get complicated, because post editing is not the same job as translating. So the term translator, I think is not future proof from that point of view that what we are asking our linguists to do is no longer pure translation. It's really a mixture of post editing and working with the machines. And working with the machines means prompting, means working on glossaries and introducing them into the workflows in being creative. These things which don't necessarily come naturally to a translator, those are things that linguists need to learn. Some are very good at that, but others have to work really hard to acquire those skills. And it's. It's really something that we are taking very seriously in terms of training our workforce to work with those new tools and going into that direction. And as I said earlier, not everyone is up for that. Not everyone's made for that. And yeah, it means that if you as a linguist can find a path and acquire that skill set and do the things I mentioned, well, I think there is a future. And conversely, if you don't feel like it, well, you know, it's time really to look for your next career step because this is not going to last for very long.
[00:26:14] Speaker A: Really like the grounded approach.
I think it's becoming more and more evident that that's the case and that this is not the only industry that's going through that. And I think that's also a big one that I've seen in other sectors and it's very important that we understand that that is across everywhere and across everywhere you have humans who adapt and augment, and that's very great.
Let's talk business facilities, your clients.
So how do you encourage adoption? Where are they hesitating? Where are they jumping right in to work with you on?
[00:26:50] Speaker B: That's a really good question, actually. And it's always a bit tricky to decide whether you want to go for early adoption or wait until a solution is really mature and appropriate. But the environment is very competitive, so, you know, just waiting and watching the market is risk in itself. So we've opted to accelerate in our development of tech solutions and approach our clients with a sense of understanding what the most important things and difficulties are that they're facing and how our solutions can help them improving the situation.
So we have launched a number of initiatives and pilot projects with our clients.
We are, rather than rolling out a product and, and telling our clients, just buy this and this is gonna help you. It's not a universal medication. It's really understanding the specific situation of the client and tweaking the solution so that it does the right thing for the client. So we offer our clients these pilots.
It's an opportunity for them to understand how the solution would work for them.
It's an opportunity for us to understand how we can improve things for the client. And hopefully at the end of the pilot, we can agree on a way forward that enables them to get better results and for us to continue to have a trustworthy relationship with our clients.
[00:28:42] Speaker A: That is a great approach. I'm going to do what Renato Baninato always says, steal the concepts. I'm going to steal that I'm going to steal. I'm going to re. Listen to this conversation and I'm going to steal those concepts. I think it's a great way to deal with solutions.
[00:28:57] Speaker B: Please don't let Renato listen in. That would be risky.
[00:29:03] Speaker A: I'm sure he's going to listen to this conversation. So. Hello, Renato, thank you for giving that perspective. If we're talking now to the localization professionals, a lot of changes are happening in that conversation. A lot of acronyms are changing, A lot of concepts are being reshaped. I hear a lot of global experience, global content, global strategy, which is very interesting. It sounds like the augmentation of something else as well. But for localization professionals, what are the skills that they need in that augmented and new environment?
[00:29:38] Speaker B: Well, I mean, that goes back to the question, what kind of services do the clients need in the Future.
And obviously, you know, language services in some shape or form will continue to be in demand, but will they be the same ones as today?
There's, you know, I think that's doubtful.
So yes, of course, you know, I mean, I think we are already, we already have seen the peak of post editing and even post editing is something that's going, I think, more downhill than uphill now.
So what are the things that linguists can do to be more future proof? I think working understanding how generative AI works is a first step and how the human can influence the LLM, I. E. How to provide relevant prompts and how to improve the performance with better prompting is a basic skill that is relatively easy to acquire, but it's really important because who is going to prompt the LLMs for the clients?
Is it going to be the clients themselves? While I don't think that clients want to burn themselves with that task, so they will ask someone else to do. And so if you're capable of doing that, great. There are other areas where linguists can play a role in anything that has to do with language and that might be data sets and labeling data and just being there when linguistic expertise is required. But I would go a bit further than all of that by saying that, yes, the term global is popping up left and right.
I think we need to go back to the basic requirement of our clients, which is they want to know how they can sell more internationally, hence the term global.
How can we help them to achieve that? It might be things like understanding the markets better for them. It's, you know, how can we provide more consultancy?
How can we ask a linguist in territory to help us with understanding the social media environment?
Can they summarize a market condition?
Can they understand their local market? And that will add much more value than just translating words.
I think if I were a linguist, I would look in those directions and try to acquire some of those skills to have something to offer in the future.
[00:32:32] Speaker A: Thank you. I think the industry is going to appreciate that and we'll probably share some of those snippets in some of our other shows. Looking ahead, what excites you about the next chapter for Alpha CRC and the localization industry as a whole?
[00:32:48] Speaker B: Yeah, I think this is one of the scariest and most exciting moments. At the same time, I think in the last decades the localization industry has seen a continuous up and up and up until maybe Covid hit and then some of the reality started kicking in. But at no point did we have an existential threat to the future of the industry. But we do have it now.
Some people believe it's stronger than and others think it's all temporary and might go away. Who knows? I can't predict really which way it's going to go. But there is an exciting thought that we can be part of a relatively limited number of companies who can shape the future and who will be left with the most valuable of all ingredients, which is this tremendous linguistic expertise that we've built up over 40 years. If we put together the linguistic expertise of all the people who work in this company, it's really amazing. Not all of them will be here in the future, but the ones who will survive and who will contribute will have a very significant role to play. And that single aspect fills me with much hope that we can be significant contributors in the future. And I think the same goes for many linguists who are taking up the challenge of remaining relevant and having something to offer.
And that's why I think combining technology and human expertise has never been a better recipe for success than today. So I'm cautiously optimistic about the future, even though I'm not trying to talk away the challenges and the threats. But there is reason to believe that there is a future for a lot of the people in this industry.
[00:34:53] Speaker A: I'm one to believe so as well.
I think a lot of innovation is going to come out of this.
I already see it happening in different places and I think it's going to get more mainstream, this capacity to innovate and create.
So I'm also very excited about the future.
[00:35:11] Speaker B: It's.
[00:35:11] Speaker A: It's almost time for us to go in this conversation. I also believe that this, this environment that, that we will have will allow for new business ideas to arise. Is. Is that part still of the unknown? Or is the same type of business just augmented?
[00:35:33] Speaker B: Yeah, No, I think it's a very relevant question because I think that the. The challenges and the threats are not coming only from the inside. And when I say inside, just the replacement of humans by the machines. But there might be solutions developed by companies who are not in this industry. And I'm not even talking about the tmss and those providers. I'm talking about companies from the outside who might come up with something groundbreaking that we had never thought of and they might become successful. I think that is a possibility and we need to keep that in mind. And we need to challenge ourselves to think outside the box and think what can be done to deliver more value to our clients.
So we can't just be on a marginal Curve of acceleration. It needs to be innovation and impactful progress that we deliver to our clients.
[00:36:37] Speaker A: Absolutely, absolutely. It's a complete revamp on. It's almost like when you get a new operating system. I think humans are getting a new operating system and we really need to get that upgrade. If you're not getting it, you're going to miss out. You're going to miss out. I definitely see it in the day to day activities, the publication and media organization.
You'd say there are not so many things to automate, but there's a lot of things to synthesize and it's, it's very helpful to use different tools to allow for that synthesis. And then it is our human intuition that allows us to understand what's not to be paid attention to at this stage.
[00:37:17] Speaker B: Well, when you think that, you know what, what makes it that machines haven't taken over humans yet? It's really the ability to define goals. The machine can't define the goals at least yet. The machine has taken over in some areas like think of medical diagnostics and because it can see things better than the human eye and it can tell you things more clearly. So in that sense, some of the tasks have been taken over by the machine. But the ability to build things to order humans and things around, that's not there yet. And as long as the machines can't acquire those things, I think the humans are still going to drive the machines. Unless of course some rogue state decides to change that.
But I think there's still hope for humanity that we will evolve and be more intelligent than our worst side and managed to guide and lead the machines rather than the other way around. For some time at least.
[00:38:27] Speaker A: I think some human beings are realizing. I've used this image before, but it's that scene in Harry Potter where he's waiting for his father to release the Patronus and then he realizes it's him, the one in all these memories. And I think some humans are starting to realize like, oh, there is a human that's going to take my job. He's like, yeah, is yourself in the future? You are in fact, yes, there is a human using AI taking your job, which is yourself in the future.
[00:38:53] Speaker B: But I think it's important to send a message to the humans in our industry, the linguists and anyone else who works in this industry, that yes, there will be jobs lost to machines, but there are also beautiful opportunities to do something more interesting tomorrow. And I would just like to encourage everyone to think about this and seek for themselves something that's interesting for the future.
[00:39:19] Speaker A: Vasilis Anas, you've said something more interesting related to language, related to culture, which is that nuance. That's an augmented nuance. Because before it was just like, oh, I'm paying you for the words. It's like, wow, you're paying me for the words and you're assuming the cultural nuance. And now I'm going to be augmented into this reality where this continues to be. This comes to light as the differentiating element of the conversation. So I'm very glad we've had this conversation. I get the feeling we could talk for hours. So hopefully in a future event, if we could talk again, I think we could continue the conversation. Before we go, I know you have sent messages to the linguists in the industry and to other professionals. Is there anything else you'd like to mention?
[00:40:04] Speaker B: I think we've covered pretty much the topic, so I would just like to thank you for the enriching conversation and hope that it's been helpful.
[00:40:19] Speaker A: Thank you, Vasilis, for your time and it's been a wonderful conversation. I'm sure we're gonna have it in the future again. And let's see if Renato mentions anything about it. Thank you for listening to Localization today. A big thank you to you again, Vasilis Hamali, for helping us explore the role of the human in localization workflows and what augmentation means for the industry's future. Catch new episodes on Spotify and Apple, podcasts and YouTube. Subscribe rate and leave a review so others can find the show. I'm Eddie Arrieta with Multilingual Media. Thanks for joining and we'll see you next time.