Lugayetu: Safeguarding the Digital Future of Low-Resource Languages in the Congo Using Artificial Intelligence

Against the backdrop of globalization, language extinction is accelerating, and the rich low-resource languages in the Congo region face the dilemma of a lack of digitalization. With the mission of 'Our Languages', the Lugayetu project uses AI technologies such as data collection, speech recognition, and machine translation, combined with a community participation model, to build digital archives for endangered low-resource languages in the Congo and develop practical AI tools. It aims to promote the protection and inheritance of linguistic diversity, allowing these languages that carry unique cultures to continue their lives in the digital age.

Low-resource languages refer to those lacking large-scale digital corpora and NLP tools; most languages in the world fall into this category. The Congo has over 200 languages, which are not only communication tools but also treasure troves of knowledge such as traditional medicine and agricultural wisdom. However, traditional AI is trained on high-resource languages, widening the digital divide. Users of low-resource languages face the risk of marginalization, and language extinction will lose unique perspectives of human civilization.

The project's core goal is to build digital infrastructure for low-resource languages:

1. Data Collection: Community crowdsourcing (contributing voice/text via mobile apps), collaboration with local institutions (digitizing written materials), oral history recording (transcribing elders' narratives);
2. Technology Development: Speech recognition (end-to-end deep learning + transfer learning/data augmentation), machine translation (multilingual NMT + unsupervised/semi-supervised learning);
3. Architecture and Open Source: Modular microservice architecture, open-source code to attract global developers;
4. Community-Led: Language ambassador network, educational cooperation to ensure native speakers participate in decision-making.

The project connects communities through a language ambassador network, coordinates data collection and feedbacks needs; collaborates with local schools to integrate language digitization into courses, enhancing young people's native language identity and acquiring data; explores sustainable business models, planning to provide AI tools to corporate clients, with revenue supporting community activities. Initial progress has been made so far, but challenges remain.

Lugayetu demonstrates the potential of AI in language protection and emphasizes the ethical responsibility of technology. The project faces challenges such as funding shortages, scarcity of interdisciplinary talents, and performance gaps in low-resource models. Future goals: Short-term—complete digitization of 10 Congolese languages and release speech/translation tools; mid-term—expand to other low-resource language regions; long-term—promote more inclusive AI development and safeguard linguistic diversity.

1. Strengthen Community Participation: Establish partnerships to ensure native speakers lead decision-making;
2. Open Source Sharing: Open data and models to build a global research community;
3. Sustainable Development: Explore commercial monetization models to ensure long-term project operation;
4. Talent Cultivation: Train interdisciplinary talents who understand technology and African languages;
5. Cross-Field Cooperation: Combine academic, corporate, and government resources to support low-resource language protection.