Studying the future
Jane Nemcova, an adjunct professor of artificial intelligence (AI) at the Middlebury Institute of International Studies (MIIS) at Monterey, teaches future linguists the ins and outs of AI’s application in the language industry. The MIIS alumna-turned-professor sees it as a way to help prepare students for a future in an industry that’s constantly evolving to reflect the latest technological innovations.
Nemcova herself has seen this evolution firsthand — having begun her career as a French linguist working in localization, she worked her way up to chief sales officer at Moravia (now RWS) and eventually served as managing director of Lionbridge AI.
With the human parity debate at the forefront of this month’s issue, MultiLingual chatted with Nemcova to hear her thoughts on how human parity in AI might change the role of future language professionals.
Editor’s note: This interview has been edited for clarity and conciseness.
MultiLingual: Let’s start with a bit about your background — could you tell me a bit about your career and how you found yourself working in the language services industry and focusing on artificial intelligence?
Jane Nemcova: Yeah, so I was a French linguist when I was starting out, and I studied business and language in France and went to the MIIS and got a master’s there. I started out in localization after I left Monterey, and started at Moravia, which is now RWS, and I was there for about 10 years and took on the role as chief sales officer.
And then Lionbridge approached me. Because I had a very purposeful strategy around technology in my years at Moravia, I could see the direction of where they were heading in deep learning and that it was a very data-intensive approach that would require new needs in the market in order to scale. And none of that was very clear to people, but to me it was kind of like the future — this is where the outsource vendors who are familiar with language are going to go, and so I started what got rebranded as Lionbridge AI.
ML: And you also teach a course at MIIS now, correct?
JN: Yeah, I do. I’m teaching an AI course for language students. I went to MIIS myself and have a great appreciation for what the students are doing and the faculty there. I thought it was a good idea to help the language students understand how to take AI and understand what was going on, be aware, and come out of it understanding the principles in artificial intelligence, and build a framework up so that when they head into the workplace, they have an understanding of it and can use their skills to the maximum there.
ML: For this February issue of MultiLingual, we’re really focusing on looking at the debate over human parity in the field of machine translation (MT). Could you offer a bit of insight into what human parity is and where we are in terms of achieving it?
JN: There are so many people who spend their careers — probably more than me — just dealing with that particular issue. And I certainly work with a lot of people who do. In terms of what I can offer, it’s this:
When I was young and doing localization work, MT was interesting, but the peripheral concept wasn’t really as important as translation management systems or something that organizes the structure of outsourcing. Machine translation created a bit of a cynicism in the localization world. [People said things] like, “Yeah, machine translation is here, but it’s not really ever going to be as good as a human.” So in the last 10 years, people’s mentality shifted toward, “OK, well, maybe MT is going to play a much more important role in this industry than we expected.”
And now, everybody’s started using MT — either customers are explicitly asking for it or vendors are using it. Now the activities [of a linguist] are shifting toward editing, and it’s a different kind of work. So, the maturity of this market, as it relates to machine translation, and the view of human parity has shifted quite a bit to, “OK, well, you’re able to get, for the most part, a lot more consistent, high-quality results than you ever thought before.”
[And with] generative AI, the AI world as a whole became a lot more educated about language. At this point, technologies like ChatGPT are demonstrating something that’s a bit of a paradigm shift for the localization world. Large language models are now a fundamental part of AI, and all these products that are going to be occurring in the future, generating content, rather than translating it. There’s probably years ahead before some of the work that’s being done today is going to get changed or modified, but it’s nevertheless looking like that’ll be more and more a reality.
I think that changes the question of human parity, because, in a way, it sort of forces one to think, “Does it even matter if you reach human parity?” If you can generate the content without a human, it kind of forces that question. There’s many dimensions you could analyze that question in and, from the linguist’s perspective, it leaves them wondering, “OK, well, what’s my future? What am I supposed to be doing? Do I have a role in language?”
And I think the answer is yes, definitely. Your ability to understand more than one language, to understand the nuance in meaning, and the way that the brain uses language as a reflection of thought is one of the most valuable skills that anybody can have — in a way it’s more valuable than engineering, it’s more valuable than one’s ability to do certain kinds of math. The stumbling block is that the people who are in that world who have those skills only envision a couple of avenues for themselves: Either you become an academic in each language, or you become a translator. And I don’t think those are the only paths.
ML: Do you anticipate that the people with language skills, who historically might have been academics or translators, might be expanding into different roles?
JN: I’m living proof of that. Nobody taught me anything about artificial intelligence initially. As a language person who was very entrenched in localization and thinking as not only a translator, but as somebody running a business, I saw that as an enormous opportunity to create new careers.
AI is forcing a different skill set — the more accuracy that we have in machine-translated content, or in any kind of AI-generated format, it’s going to require humans who have skill in language to play a role in that. They’re also going to be required to play bigger roles from a number of different standpoints, including ethics and bias.
I think linguists and people who understand the structure of language as it relates to our own cognitive abilities have a very important role to play in that conversation, and how to manage AI, how to improve it, debias it, and resolve ethics questions that are fundamental to a product delivery to the market. That’s really needed to protect consumers, the management, and the staff.
Once again, we’re reaching a stage where technology is happening so quickly that the market and governments are catching up to it people are trying to figure out, it’s like an information overload that people are coming to terms with. But I think that it’s only going to mean that people with language skills are going to be more needed in the future rather than less.
How to Future Proof Today’s Machine Translation
Did you know that the amount of content produced by humanity each day is estimated to be 2.5 quintillion bytes? That's more bytes of text…→ Continue Reading
For the Best Translations Humanity is Key
Three decades ago, when I started working as a professional translator, I sent documents to clients by post . . . and then waited days…→ Continue Reading
Subscribe to stay updated between magazine issues.