The Team Building AI to Help Africa Fight Misinformation
Improvements in generative artificial intelligence are being made every day but misinformation is on the rise too. Tools developed and trained in Africa by Africans are now assisting journalists and the general public in distinguishing between true and fake information. Monsur Hussain and his team are at the forefront of a quest to check fake info at the door.
In Abuja, Nigeria, Monsur Hussain, the innovation head at the Centre for Journalism Innovation and Development (CJID), leads a team of software developers and AI researchers focused on improving fact-checking technology. The team is developing Dubawa AI, a fact-checking tool launched in 2024, to help journalists and the public verify claims before they spread online. Hussain emphasizes the importance of understanding claims, particularly those with societal impact such as health and political claims.
Nigeria has over 148 million active internet subscriptions, making it Africa's largest online market. However, generative AI has made it more challenging for journalists to verify content due to convincing deepfakes. Local AI models are crucial for African journalists as they are trained on regional information, reducing the risk of "hallucinations" often seen in global models.
Japhet Johnson, a software engineer based in Abuja, stresses the need for AI systems that understand the context of African news and comply with local data protection laws.
Dubawa AI includes a chatbot and an audio tool that allows users to check claims within seconds. The chatbot, accessible via dubawa.org, has verified over 6,000 claims and is used in more than 32 African countries. The development process involved creating a minimum viable product, gathering feedback from journalists and fact-checkers, and iterating until a satisfactory solution was achieved.
A study found that 75% of Nigerian internet users frequently encounter fake news on social media platforms. Hussain believes that AI can help reduce misinformation and monitor claims more efficiently, while still allowing for editorial judgment. The team implemented safeguards to prevent the AI model from going off-topic, ensuring reliable performance. Hussain views Dubawa AI as a step towards accountability in the dissemination of information.
Written by urgent.news from Capital FM Kenya's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
