Responsible AI for Low-Resource Languages
Building language tools that respect context, consent, and community ownership.
Beyond English
Most modern AI is trained on a tiny slice of the world's languages. For Africa's hundreds of languages, data scarcity is the core challenge.
Principles
- Community ownership of language data
- Transparent provenance
- Security and consent by default
We are assembling openly-licensed corpora and benchmarks so that researchers can build tools with communities, not extract from them.
nlplanguagesethics