A mobile phone is all you need. Just enter a sequence of numbers and anyone in Rwanda can access a chatbot in Kinyarwanda, the official language, that provides the latest coronavirus information. How many cases are there? Where can I be tested? What do I need to do when travelling? Having a service like this in a country’s official language is by no means a given. English, French, Spanish and German have been the main languages used by digital voice assistants to date. That excludes millions of people. African and Asian languages have been underrepresented until now in the development of language-based artificial intelligence. And that makes things all the more difficult for women and men who cannot read or write.
In 2020, GIZ worked in Rwanda with the Mozilla Foundation, which belongs to browser developer Mozilla, and with local partners, including the start-up Digital Umuganda, to record and process more than 2,000 hours of content in Kinyarwanda. The datasets are openly available to all developers at no charge. They have been used to create a prototype chatbot, a text-based dialogue system. This development of local AI is an example of FAIR Forward – Artificial Intelligence for All, an initiative of the German Development Ministry that GIZ is implementing. It aims for a more open, inclusive and sustainable approach to artificial intelligence at international level. The primary objective is to strengthen the basis for locally developed AI in selected partner countries, so that digital innovations drive forward efforts to achieve the Sustainable Development Goals. With this focus, the global project is therefore also implementing the objectives of the German Government’s Artificial Intelligence Strategy, and is currently operating in Ghana, India, Rwanda, South Africa and Uganda.
Free access to knowledge for 12 million people
The East African nation of Rwanda is one of the pioneers. ‘Kinyarwanda has become the fastest-growing dataset and second-largest open voice dataset in the world,’ reports Audace Niyonkuru, founder of Digital Umuganda. Umuganda is the name given to the one day in the month on which everyone in Rwanda carries out community work. The start-up has brought this day into the digital age and regularly calls on people to ‘donate’ their language so that others can benefit. The sentences in Kinyarwanda have been recorded by thousands of volunteers. The language is spoken by more than 12 million people, not only in Rwanda itself but also in neighbouring countries.
The GIZ initiative is working with Mozilla and other partner organisations including universities. Gathering the language data is just the start. Openly available AI training data can be used to strengthen the entire local digital environment. It allows developers from other regions to access the data too and adapt the information for their own, local purposes. The data gathered is incredibly versatile and can be used in areas as diverse as interactive citizen participation, apps that identify plant varieties and diseases, and chatbots that can answer questions on sustainable agriculture. This open form of data gathering fosters digital innovation, particularly in the Global South, and also plays a role in democratising artificial intelligence.
Sustainable Development Goals (SDGs)
The Sustainable Development Goals (SDGs) in the 2030 Agenda provide the framework for our activities around the world. Click on a goal to find out which projects in this report are helping to achieve it.