Highlights from the TAUS Massively Multilingual AI Conference in Albuquerque

The TAUS Massively Multilingual AI Conference in Albuquerque provided a platform for (as always) thought-provoking discussions, but this time was on how AI is reshaping language services and the role of human translators. 

This conference did not only have one of our best programs ever, but it also marked the celebration of 20 years of TAUS, close to the birthplace of TAUS data in Taos, NM. This article is a summary of the takeaways that were shared about the event by attendees.

The Evolving Role of Human Translators

As Marina Pantcheva, PhD wrote in her article earlier, AI has drastically changed the translation landscape. Manuel Herranz ( Pangeanic GenAI for Language Technologies and Services ) noted, language is becoming less human-centric. This shift means that traditional translation services now serve more as "insurance" for language service providers (LSPs).

Jaroslaw (Jarek) Kutylowski ( DeepL ) emphasized that human translators are evolving from “doing” the translations to overseeing AI-generated content.

Tomorrow's translator will be a supervisor, linguist, data analyst, and prompt designer. But as Marco Trombetti ( Translated ) pointed out that AI may eventually surpass human intelligence, though humans are still critical to the process—for now.

Oscar Straker 's demonstration of personalized AI for speech impairments and Scott S. ’s ( Amazon Web Services (AWS) ) AI-powered digital humans for sign language translation underscore the potential for AI to assist those with physical disabilities. 

Olga Beregovaya ( Smartling ) also shared how South African children are using AI as educational assistants, showing AI's far-reaching impact across demographics.

Kalika Bali (Microsoft) underscored the urgency of preserving underrepresented languages through multilingual AI. Without digital representation, languages risk extinction.

Kirti Vashee ( Translated ) and Adam Youngfield ( The Church of Jesus Christ of Latter-day Saints ) highlighted the challenges in AI performance beyond the top 30 languages, stressing the need for linguistic expertise to improve AI for low-resource languages.

The Importance of Quality Data

Speakers at the conference emphasized the growing importance of quality data over sheer quantity in AI training. Synthetic data, if used correctly, can fill gaps, particularly for low-resource languages, but human oversight is essential to maintain ethical and culturally sensitive AI outputs.

Speech Recognition and Translation Innovations

Advances like Whisper have reduced costs in speech recognition, while multi-layer AI models are showing promise in areas like transcreation and speech synthesis. However, scaling AI effectively across diverse languages, including sign language, remains a challenge.

Contest Spotlight

One of the conference highlights was the contest hosted by Anne-Maj van der Meer and Veronica Hylák ( Metalinguist ). 

Andrew Miller from BLEND Localization took home the win, showcasing how the company developed an end-to-end solution using the TAUS Estimate API that streamlines translation workflows by reducing costs and turnaround times.

The solution leverages carefully selected QE thresholds to automatically approve MT strings that meet predefined standards, allowing them to bypass human review. This approach has already been implemented by several clients, resulting in significant cost savings.

This is what Andrew had to say: 

"Attending my first TAUS event, I was blown away by the depth of knowledge, experience and quality of the discussions. Fantastic event to focus on our industry's challenges and solutions. I ran a Quality Estimation Program Workshop and presented BLEND's end-to-end MTQEPE solution in the Innovation Challenge, which won first place! A huge thank you to the TAUS team and BLEND R&D for execution!"

Networking Dinner

The TAUS Massively Multilingual AI Conference also featured a memorable networking dinner at El Pinto restaurant, sponsored by Translated . 

Attendees enjoyed engaging conversations, a mariachi band, and a relaxed atmosphere with authentic cuisine.

Future Outlook

While AI is undoubtedly transforming the industry, there’s hope for human translators. As John Fennelly ( Lionbridge ) noted, the translation market hides latent demand, and AI may free up the next generation to focus on more complex tasks like generating edge case data for machine learning.

Innovation is key to staying ahead, and as Wayne Bourland ( Dell Technologies ) pointed out, AI isn’t just about adding data scientists to the team but fostering curiosity and creativity within the entire R&D process.

A heartfelt thank you to everyone who attended the TAUS Massively Multilingual AI Conference. Your participation, insights, and knowledge made this event truly exceptional. 

We look forward to continuing these conversations and driving innovation in the future of language services together, so be sure to stay tuned for updates on the TAUS 2025 events agenda!

With AI and human expertise working hand in hand, the future of language services is brighter than ever.