Meta and UNESCO Partner to Enhance AI Translation for Underserved Languages

Meta's Ambitious AI Translation Initiative

Meta, in partnership with UNESCO, has launched a groundbreaking program aimed at improving AI-driven translation and speech recognition technologies, with a particular focus on underserved and indigenous languages 1

1

2

2

. This initiative, known as the Language Technology Partner Program, seeks to address the linguistic diversity challenges in AI development and support UNESCO's work during the International Decade of Indigenous Languages 3

3

.

Collaboration and Data Collection

The program is actively seeking collaborators who can contribute substantial language resources, including:

• Over 10 hours of speech recordings with transcriptions
• Large amounts of written text (200+ sentences)
• Sets of translated sentences in diverse languages 2

  2

These contributions will be integrated into AI speech recognition and translation models, which will ultimately be open-sourced and made freely available to the community 3

3

.

Notable Partnerships and Focus Areas

The government of Nunavut, a northern Canadian territory, has already joined the initiative as a partner. Their participation aims to develop translation systems for Inuit languages, specifically Inuktitut and Inuinnaqtun 1

1

4

4

. This collaboration underscores Meta's commitment to supporting underserved languages and aligns with UNESCO's efforts in preserving and promoting indigenous languages.

Open-Source Translation Benchmark

Complementing the Language Technology Partner Program, Meta is introducing an open-source translation benchmark called BOUQuET 1

1

. This standardized test, composed of sentences crafted by linguistic experts, will evaluate the performance of AI translation models 2

2

. Currently supporting seven languages, the benchmark is accessible on the AI development platform Hugging Face, encouraging community contributions and fostering collaborative improvement of translation technologies 2

2

5

5

.

Meta's Ongoing Commitment to Language Technology

This new initiative builds upon Meta's previous efforts in the field of language technology:

1. The No Language Left Behind (NLLB) project, released in 2022, which provided an open-source machine translation engine for numerous languages 3

   3

   5

   5

   .
2. The Meta Massively Multilingual Speech (MMS) project, which scaled speech recognition to over 1,100 languages and introduced zero-shot transcription capabilities 3

   3

   5

   5

   .

Implications and Future Prospects

Meta's investment in multilingual AI has significant implications for language support, accessibility, and the development of adaptive AI systems 5

5

. By focusing on underserved languages, the company aims to create more inclusive technology that can understand and respond to complex human needs across diverse linguistic and cultural backgrounds 2

2

5

5

.

The Language Technology Partner Program is accepting applications until March 7, 2025, with next steps to be discussed by April 15, 2025 4

4

. As Meta continues to expand its AI capabilities, including plans to significantly increase its GPU capacity by the end of 2025, these language initiatives are poised to play a crucial role in the company's broader AI strategy 4

4

.

Challenges and Criticisms

While Meta frames these initiatives as philanthropic, it's worth noting that the company stands to benefit from improved speech recognition and translation models across its platforms 2

2

. Additionally, Meta has faced criticism for its handling of content in non-English languages, with reports highlighting discrepancies in content moderation and misinformation flagging across different languages 2

2

.

As Meta advances its language technology efforts, the tech giant will need to address these concerns and demonstrate a commitment to equitable language support across its diverse user base.