Key facts
- Nonprofit Current AI is building open, public AI infrastructure to address the lack of representation for many languages and cultures in current AI systems.
- The organization aims to create an alternative to proprietary AI systems, akin to the World Wide Web, making AI accessible and free for all.
- Current AI has secured $400 million in funding, seeded by the French government, with contributions from foundations and companies.
- Recent initiatives include Suno Sutra, an offline AI device for 22 Indian languages, and Alpha Chat, an open-source chatbot developed by a coalition of organizations.
- Grants totaling $3.2 million have been awarded to projects in Kenya, Lebanon, and the Brazilian Amazon focused on AI datasets, cultural history digitization, and offline AI tools for Indigenous communities.
- Current AI emphasizes local data storage, community involvement, and consent protocols to ensure data ownership remains with communities, not tech companies.
The nonprofit Current AI is working to create an open and accessible AI infrastructure, aiming to provide a public alternative to proprietary systems developed by major tech companies. CEO Ayah Bdeir emphasized the need for AI to represent the world's diverse languages and cultures, noting that current large language models are predominantly driven by English, leaving many communities behind.
Current AI has secured significant funding, totaling $400 million, with an initial $100 million from the French government and contributions from foundations like Ford and MacArthur, as well as companies like DeepMind and Salesforce. The organization's strategy involves fostering public-private partnerships to fund public interest technology.
Recent initiatives include the development of Suno Sutra, a pocket-sized, offline AI device capable of running in 22 Indian languages, created in collaboration with India's Bhashini. This device addresses the challenge of AI's limited representation of India's linguistic diversity. Additionally, Current AI launched Alpha Chat, an open-source chatbot developed in collaboration with ten organizations, including Hugging Face, Mozilla, and MIT Media Lab.
The organization has also awarded $3.2 million in grants to four projects focused on AI development in underserved regions. These include building AI datasets for over 50 African languages in Kenya, digitizing Arab cultural history in Lebanon, developing offline AI tools with Indigenous communities in the Brazilian Amazon, and creating AI accountability tools for Africa. Current AI prioritizes local data storage and community control over data, contrasting with the market-driven approaches of Big Tech.