A New Framework to Help Every Language Thrive in AI

Artificial intelligence is changing how people search for information, learn new skills, access services, and communicate with one another. But not every language is being supported by the AI technologies people are increasingly relying on. Many risk being left behind entirely due to a lack of language data.

Today, the Coalition on Digital Impact (CODI) is proud to release a new report that offers a practical framework to change that. 

The report is the culmination of months of work by the Minimum Viable Dataset (MVD) Working Group, convened as part of CODI’s first research initiative focused on advancing multilingual AI. This group of experts in artificial intelligence, linguistics, digital access, Indigenous data governance, and cultural preservation developed this framework to answer one critical question: What does a language need to participate meaningfully in the age of AI?

‍

Translation is only the beginning

The report’s key findings reflect the need to change how people think about language inclusion in AI. 

• Translation is not representation. AI can translate a language without representing the people, culture, and knowledge behind it.
• Every language deserves a path into the AI era. The report gives communities a practical roadmap to assess where they stand and decide what comes next. 
• Technology alone isn’t enough. Meaningful language inclusion requires technical readiness, cultural representation, and community governance working together.

‍

Resources for communities

The Working Group’s report introduces recommendations, guidelines and practical tools for determining what makes a dataset culturally relevant and ethically sourced for use in generative AI models. In other words, they’ve researched what it takes to prevent a language from disappearing in an AI-enabled world. Rather than telling communities what they should build, the report gives them a flexible framework they can use to define their own goals.

The framework helps communities:

• Understand how well their language is currently represented in AI
• Identify what data is needed to support different kinds of AI tools
• Ensure cultural knowledge is represented accurately
• Protect community ownership, consent, and governance over language data
• Build a step-by-step roadmap that reflects their own priorities

Whether the goal is making a language searchable online, creating translation tools, or eventually supporting advanced AI applications, communities can choose the path that makes sense for them.

‍

Looking ahead

This report marks the first phase of CODI's work to advance multilingual AI. The next phase will focus on practical implementation, transforming the framework into a usable toolkit for language communities. If you would like to collaborate with CODI, please contact us at info@codi.global. 

As AI continues to shape how people experience the Internet, ensuring every language has the opportunity to participate is more important than ever. 

Today, we invite you to explore the report, use the framework to assess where your language stands, and share what you learn as you put it into practice. Together, we can help build an AI future where every language belongs.