Many people in the language service industry are anxious about how artificial intelligence (AI) will impact our field, but personally, I’m more excited than scared.
I’m quite lucky to work with AI on a regular basis in my role as a project manager at Intento, Inc., based in San Francisco. We are focusing on configuring AI agents to meet the range of requirements our customers have for translating various types of content across different touchpoints for their customers and employees. I see that AI has a big role to play for companies, translators, and end users. While it will change the way we all work, I’m excited about the potential.
Linguists are asked to do a lot of things when they need to produce a translation, including fixing errors in terminology, tone of voice, and gender forms, and making sure that the translation conforms to the style guide, while also being asked to bring that all-important cultural nuance and context to make the translation authentic for the reader or end user. However, the reality is that linguists don’t get unlimited time to produce a translation and therefore often have to focus primarily on fixing the errors that exist, at the expense of focusing on cultural nuance and context. That’s a loss as that is what makes translation, as a job, rewarding.
We’ve found that our customers want to use AI agents to automate those error-fixing tasks so they can let their linguists focus on making the translations perfect.
A lot of my knowledge from the Translation Technology and Website Translation classes that I took at the Middlebury Institute have come into play in understanding the more technical aspects. As our language engineers explain the technical solutions we can provide to clients, it’s important as a project manager to take what they say and help simplify it into less technical, easier-to-digest terms. Being exposed to what goes into training engines with various corpora (including doing a whole training and tuning of a project in my localization master’s program) has helped give me the tools to explain how large language models (LLMs) can be effective for tasks like linguistic quality assurance (LQA) when training an engine to see how well the engine performs after customization.
Overall, as AI is doing a lot of the work, it also means translators can produce more content, more quickly, with the same human touch, and the end users are getting a better experience as a result! I’m thankful to the Middlebury Institute for giving me hands-on experience that I would not have gotten elsewhere so that I feel more confident in my job as a PM at a localization tech company.
I recently moderated a webinar my company hosted called “Debunking Myths about LLMs and Machine Translation” with Andrzej Zydroń, CIO and cofounder of XTM International and Grigory Sapunov, CTO and cofounder of Intento, Inc.
The biggest takeaways for me were how LLMs can handle more specialized content. Machine translation (MT) can be quite literal, and we see LLMs more often being used on creative copy. As I get to manage more projects where an LLM engine wins over a neural machine translation (NMT) engine, I also get to see how LLM-based glossaries are applied in translation workflows, something I didn’t know was possible before this webinar!
This piece is part of an ongoing series featuring perspectives from our faculty, alumni, and students on the future of language services in the age of AI.
Keep up with the latest from our Translation and Localization Management Program on their LinkedIn page.
| by Caitlin Fillmore
Attendees discuss current job openings and internship opportunities with employers like Stanford Health Care, TransPerfect, Honda, and the U.S. State Department, making connections with alumni working there now.
Translation and Localization Management graduate Autumn Smith shares how her coursework and active networking helped her launch a career in localization.
| by June Jiang MATLM ’24
Translation and Localization Management graduate June Jiang shares how her coursework and the Institute’s alumni network helped her launch a career in localization.
