AI can now translate languages you’ve never heard of—Meta’s groundbreaking project revealed

Meta Platforms Inc. has unveiled its ambitious Language Technology Partner Program, an initiative designed to advance AI translation technologies and support the preservation of underserved languages. This strategic program aligns with the efforts of the United Nations Educational, Scientific and Cultural Organization (UNESCO) to promote linguistic diversity inclusivity within the digital landscape. By expanding the capabilities of artificial intelligence (AI) models to accommodate diverse languages, Meta aims to bridge the digital divide and foster global communication.

How is Meta addressing linguistic diversity with AI?

Meta’s Fundamental AI Research (FAIR) team is at the forefront of this initiative, focusing on developing advanced machine intelligence (AMI) capable of performing complex cognitive tasks such as real-time translation. By enhancing AI translation technologies, the company is pushing the boundaries of what AI can achieve in terms of multilingual support.

The FAIR team’s vision goes beyond technical advancement; it’s about ensuring that no language—and by extension, no community—is left behind in the digital age. Meta’s partnership with UNESCO under the International Decade of Indigenous Languages highlights its dedication to preserving linguistic heritage while promoting digital inclusion. This initiative is vital in creating intelligent systems that not only process but also understand languages with cultural context, supporting linguistic diversity inclusivity.

What is the Language Technology Partner Program?

Meta’s Language Technology Partner Program is a collaborative framework inviting organisations, linguistic researchers, and local communities to contribute to the development of inclusive AI models. The program’s primary goal is to improve AI translation technologies through real-world linguistic data.

Participants are encouraged to contribute:

• 10+ hours of speech recordings with accurate transcriptions
• 200+ written sentences in various languages
• Translated sentence datasets for AI training

These resources will help Meta’s AI models better recognise and translate languages that currently lack robust digital representation. In exchange, partners gain exclusive access to technical workshops conducted by Meta’s research teams. These sessions offer insights into leveraging Meta’s open-source models for creating impactful language technologies.

A standout collaboration includes the Government of Nunavut, Canada, which will contribute linguistic data in Inuktitut and Inuinnaqtun, two indigenous languages from the Arctic region. This partnership reflects Meta’s broader commitment to supporting languages at risk of digital extinction, a crucial step towards linguistic diversity inclusivity.

Why is open-source translation benchmarking important?

In addition to the partner program, Meta is launching an open-source machine translation benchmark to evaluate the effectiveness of AI models in translation tasks. This benchmark consists of carefully crafted sentences by linguistic experts, representing a wide range of languages and dialects.

Meta’s benchmark serves as a standard tool for assessing the quality and accuracy of AI translation models. By making this resource open source, the company encourages global contributions that will enrich the dataset and enhance the performance of AI translation technologies. Available in seven languages, the benchmark’s multilingual nature showcases Meta’s dedication to linguistic diversity inclusivity and the continuous improvement of language models.

Meta’s ongoing commitment to underserved languages

Meta’s dedication to language inclusivity is rooted in its history of groundbreaking projects. In 2022, the company launched the No Language Left Behind (NLLB) project, a neural machine translation engine designed to support languages previously neglected by mainstream AI technologies. This project laid the foundation for future advancements in AI translation technologies.

Building on this success, Meta collaborated with UNESCO and Hugging Face to develop a language translator based on NLLB. This partnership was showcased during the United Nations General Assembly in 2023, further solidifying Meta’s role in promoting linguistic diversity inclusivity on a global stage.

Meta’s Massively Multilingual Speech (MMS) project, another key milestone, has expanded audio transcription capabilities to over 1,100 languages. In 2024, the project introduced zero-shot speech recognition, enabling AI systems to transcribe languages they have never encountered before—without prior training. This innovation represents a significant leap in the evolution of AI translation technologies, allowing for real-time adaptation to new linguistic environments.

What’s next for AI translation technologies at Meta?

Meta’s long-term vision is to develop intelligent systems capable of understanding and responding to complex human communication needs, regardless of linguistic or cultural backgrounds. By fostering partnerships through the Language Technology Partner Program, Meta aims to co-create solutions that reflect the world’s rich linguistic tapestry.

As Meta continues to refine its AI translation technologies, its focus remains on collaboration, open-source innovation, and the promotion of linguistic diversity inclusivity. The company invites interested parties to join this transformative journey by applying through its online interest form, contributing to a future where every language has a voice in the digital world.