Introduction
The Africa Infodemic Response Alliance(AIRA) is a collaborative initiative led by the World Health Organization(WHO) that was launched in 2020. It brings together inter-governmental and non-governmental organisations working in public health, as well as fact-checking and media organisations, to tackle the problem of infodemics.
As WHO states, an infodemic occurs when there is an overwhelming amount of information, including false or misleading details, during a public health emergency. AIRA works to combat the spread of false information and provides accurate and timely health data to its members, partners, and communities across Africa.
AIRA's key activities include listening to narratives online and offline, providing technical support to African countries to strengthen their information ecosystems, producing health literacy, debunking and prebunking social media content, and continuously monitoring the impact of its work in combating health information. AIRA utilises AI for monitoring and analysing vast amounts of data from online (news articles, social media including WhatsApp) and community-based sources to identify misinformation trends and public concerns.
Background
Managing infodemics, especially about health, is a critical challenge in the humanitarian sector. Information disorders during health crises can severely impact public safety and well-being, particularly where reliable information is scarce. For example, the COVID-19 pandemic showed how rapidly false information can spread, creating fear, confusion, and hindering effective public health responses, as noted by the WHO. This wave of misleading information undermines trust in health systems and can lead to dangerous behaviors, such as rejecting vaccinations or endorsing unproven treatments.
To tackle this, AIRA uses AI to support social and community listening (social listening or social and community listening refers here to the identification and analysis of narratives on a specific topic that are shared on online platforms and also discussed in communities). This allows for the efficient identification and analysis of health conversations to understand public opinion and sentiment. AI's ability to sift through vast amounts of data quickly and accurately helps identify information patterns that would be challenging for humans to detect manually. The platform also allows the inclusion of audio, videos, WhatsApp voice messages and text messages which is unprecedented. Moreover, the platform uses AI to translate content in more than 200 languages, and it can generate short analytical reports. Finally, a catalog of WHO-cleared prebunks and debunks is made available to users to quickly counteract the harmful narratives identified.
By effectively identifying and countering infodemics in real-time, AIRA ensures that accurate health information reaches communities, helping them make evidence-based decisions and build trust in health systems.
AI development
AIRA leverages artificial intelligence through their infodemic management framework, which comprises four pillars: identifying information gaps and misinformation, simplifying technical knowledge, amplifying correct information, and quantifying the impact of interventions. AI is employed in the first two pillars.
Identify: The AIRA team uses AI-assisted social media listening tools such as NewsWhip Spike, Meltwater, Google Trends and Google Alerts. These tools use AI to track and analyse vast amounts of data from online platforms. For example, NewsWhip Spike uses AI to monitor trending news stories and social media posts in real-time, helping to spot emerging topics and potential misinformation. Meltwater uses AI to collect and analyze online content from various sources. Google Alerts uses AI to send notifications about the latest content matching specific keywords, keeping the team updated on new articles and social media posts related to public health topics, and Google Trends uses AI to analyse search patterns and provide insights into trending topics, allowing the team to stay informed about the latest public health issues and track changes in interest over time. By leveraging AI, these tools help gather both qualitative insights (people's opinions and feelings) and quantitative data (number of mentions and trends), allowing the team to pinpoint misinformation trends and understand public concerns more effectively.
Simplify: To make complex health information as accessible as possible, the team uses the Africa Misinformation Portal (AMP), a web-based platform developed by AIRA. AMP curates information, labels it using AIRA’s specialised health taxonomy, translates it into multiple languages, geolocalises data, and generates basic analytical reports to identify trends and patterns.
Additionally, AMP ensures that accurate and verified health data is widely shared through social media, websites, newsletters, and partnerships with local community organisations. This approach ensures that correct health information reaches a broad audience, effectively combating infodemics.
Operationalising AI
AIRA operates across Francophone, Lusophone and Anglophone African countries. The work includes the distribution of weekly AIRA Infodemic Trends Reports and real-time alerts of concerning narratives. These reports and alerts are shared primarily with AIRA members such as UNICEF, Africa CDC, and participating and supporting entities such as AFP fact-checkers and Africa Check. When misinformation or concerning trends are identified, AIRA ensures that these key organisations are promptly informed to address these issues.
The primary challenge AIRA has seen involves capturing contextual nuances, collecting timely information from offline channels and from closed messaging apps (such as WhatsApp, Telegram). The team therefore combines the automatically collected data with information provided by local partners in countries, including Ministry of Health focal points, infodemic managers, and fact-checkers. This triangulation of data sources provides comprehensive and accurate reporting by identifying the subtleties and local context in infodemic trends.
Additionally, AIRA works on reducing potential biases introduced by the social listening analysts themselves. The team has implemented a manual review process to validate the data and works with community-based individuals to gain insights related to language, cultural context, and other local factors to ensure that the analysis and information provided by AIRA is both accurate and context aware.
Learnings
AIRA’s overall verdict on the use of AI tools is that there are significant benefits. The efficiency of efficiency in processing and analysing large volumes of data from social media listening is vastly improved. AI tools enable quick analysis and pattern identification, allowing for more effective responses to misinformation. However, the team also highlights the importance of the human element. While AI tools can handle vast volumes of data, human analysts are essential for contextual understanding and verification; and the manual review process ensures that the data is accurately interpreted. Besides, the creation of an AI-assisted tool is not enough. AIRA is exploring social behavior change techniques to understand barriers and promote a wider adoption of the tool in the user’s habits.
Plans for the future
AIRA's plan focuses on strengthening networks and standardising data collection at the country level to ensure more comprehensive and accurate reporting. By having more data and a standardised approach, AIRA aims to further improve both the efficiency and quality of their data collection efforts. To facilitate even more detailed and localized data collection, better understand region-specific misinformation trends, and to ensure the AI analysis is complemented with in depth understanding of language, culture, and local context, there will be two separate coordination teams in the future who will be focusing on monitoring in Anglophone and Francophone regions respectively. AIRA also plans to conduct webinars and provide guidance on the use of AI in social listening to enhance the network’s capabilities and increase the use of the information and analysis provided.
Where to learn more
Elodie Ho, AIRA Coordinator, hoelodie@who.int The Africa Infodemic Response Alliance