Meta, the tech giant behind Facebook and Instagram, has announced a new partnership with UNESCO to launch the Language Technology Partner Program, a initiative aimed at collecting speech recordings and transcriptions to develop open-source AI speech recognition and translation models. The program seeks to promote language diversity and inclusivity by focusing on underserved languages, according to a blog post provided to TechCrunch.

The Language Technology Partner Program is calling on collaborators to contribute more than 10 hours of speech recordings with transcriptions, large amounts of written text, and sets of translated sentences in diverse languages. These contributions will be integrated into AI speech recognition and translation models, which will be open-sourced once finalized. The government of Nunavut, a sparsely populated territory in Northern Canada, is one of the first partners to join the program, with a focus on supporting the Inuktut languages spoken by some of its residents.

Meta's efforts are driven by the goal of creating intelligent systems that can understand and respond to complex human needs, regardless of language or cultural background. The company's AI teams will work closely with partners to develop AI models that can accurately recognize and translate speech in various languages. This initiative is particularly significant, given the criticism Meta has faced in the past for its handling of content in languages other than English across its platforms.

In a related move, Meta is releasing an open-source machine translation benchmark to evaluate the performance of language translation models. The benchmark, composed of sentences crafted by linguists, currently supports seven languages and can be accessed and contributed to on the AI development platform Hugging Face. This benchmark will provide a standardized way to measure the accuracy of language translation models, promoting further innovation and improvement in the field.

While Meta is framing these initiatives as philanthropic, the company stands to benefit from upgraded speech recognition and translation models. Meta continues to expand the number of languages its AI-powered assistant, Meta AI, supports, and is piloting features such as automatic translation for creators. Last September, the company announced that it would begin testing a tool to translate voices in Instagram Reels, allowing creators to dub their speech and auto-lip-sync it.

Meta's treatment of content in languages other than English has been a subject of controversy, with reports suggesting that the company has been inconsistent in its moderation of misinformation and hate speech across different languages. According to one report, Facebook left almost 70% of Italian- and Spanish-language COVID misinformation unflagged compared to just 29% of similar English-language misinformation. Leaked documents have also revealed that Arabic-language posts are regularly flagged erroneously as hate speech.

Despite these challenges, Meta has committed to improving its translation and moderation technologies. The Language Technology Partner Program and the open-source machine translation benchmark are significant steps towards achieving this goal. By promoting language diversity and inclusivity, Meta aims to create more equitable and accessible AI systems that can benefit people from diverse linguistic and cultural backgrounds.

In the broader context, Meta's initiatives highlight the importance of language diversity and inclusivity in the development of AI systems. As AI becomes increasingly pervasive in our daily lives, it is essential that these systems can understand and respond to the complex needs of people from diverse linguistic and cultural backgrounds. By promoting open-source collaboration and innovation, Meta's initiatives have the potential to drive meaningful progress in this area and create more inclusive and equitable AI systems for all.