Hajra Naz
Meta has announced the launch of a new initiative, the Language Technology Partner Program, in collaboration with UNESCO. This program aims to collect speech recordings and transcriptions to enhance the development of future AI models that will be openly available.
The initiative seeks partners who can contribute over 10 hours of speech recordings with transcriptions, large datasets of written text, and extensive sets of translated sentences in varied and underrepresented languages. These partners will work closely with Meta’s AI teams to integrate these languages into translation and speech recognition models, the company said. These AI models will be open-sourced once they are created, allowing academics and developers all across the world to use them.
The government of the inadequately possessed Northern Canadian territory of Nunavut is one of the unique accomplices. Inuktut is one of the Inuit languages spoken by numerous individuals in Nunavut. Meta trusts to encourage UNESCO’s objective of protecting linguistic differences by joining these underrepresented languages.
In a blog post, Meta highlighted the program’s focus on underserved languages. The company stated:
“Our efforts are especially focused on underserved languages, in support of UNESCO’s work. Ultimately, our goal is to create intelligent systems that can understand and respond to complex human needs, regardless of language or cultural background.”
Read More: No Fact-Checks, No Limits: Meta’s New Internet Philosophy
Meta’s Open-Source AI Translation Benchmark
As a complementary effort to this new program, Meta has also announced the release of an open-source machine translation benchmark. This benchmark, designed to evaluate the performance of AI-powered language translation models, consists of carefully crafted sentences created by linguists. It currently supports seven languages and is available on the Hugging Face AI development platform, where researchers and developers can both access and contribute to the dataset.
While Meta presents these initiatives as philanthropic efforts, the company also stands to gain significantly. By enhancing its AI-driven speech recognition and translation capabilities, Meta can further expand the reach of Meta AI, its AI-powered assistant. Additionally, this will bolster AI-powered automatic translation tools for creators across its platforms.
AI-Powered Language Features on Meta’s Platforms
Meta has already been incorporating AI-powered language tools across its ecosystem. Last September, the company introduced an automatic voice translation feature for Instagram Reels, allowing creators to dub their speech and auto-lip-sync in different languages. These advancements position Meta as a leader in AI-driven multilingual content creation and accessibility.
However, Meta’s history with non-English content moderation has sparked controversy. A past report revealed that nearly 70% of Italian- and Spanish-language COVID misinformation remained unflagged, whereas only 29% of similar English-language content was left unmoderated. Additionally, leaked internal documents showed that Arabic-language posts were frequently misclassified as hate speech, leading to errors in content moderation.
Meta’s Commitment to Improving AI Translation and Moderation
Meta has acknowledged these difficulties and said it is making a concerted effort to enhance its moderation and translation tools. The company’s goal is to provide more responsive and inclusive digital experiences for people globally by emphasizing language variety and AI-powered accuracy.
Meta is making significant progress in using AI to overcome language barriers through its collaboration with UNESCO and the Language Technology Partner Program. It is unclear if this endeavor can successfully rectify earlier issues with AI-driven translation and moderation, but it does represent Meta’s continued investment in AI for international communication.
