Citizen Science and AI: A Path to More Effective Poverty Alleviation
The application of artificial intelligence (AI) to humanitarian efforts is often overlooked, despite its great potential. A recent example is in the use of AI in Togo during the COVID-19 pandemic. The Togolese government leveraged AI to identify tens of thousands of households in need of financial assistance. This was a significant breakthrough, as traditional methods like welfare applications and household surveys were impossible during the pandemic.
This innovative approach involved using machine learning to analyze satellite imagery of low-income areas, alongside data from mobile-phone networks. This allowed authorities to identify eligible recipients and provide them with regular payments through their phones, demonstrating the transformative power of AI in addressing immediate needs.
Volunteer researchers have been collecting litter data on Ghana’s coast, which the government is now including in its official statistics on the environment.
The Challenge of Data Collection in Poverty Research
Post-pandemic, researchers and policymakers are exploring how AI can improve poverty alleviation strategies. However, effective poverty reduction demands comprehensive and reliable data on household conditions. Crucial details such as housing quality, children’s diets, education, and healthcare access are necessary to understand and address poverty in the most effective way. Traditional in-person surveys are typically used to gather this vital information.
However, gathering such data, particularly in low- and middle-income countries (LMICs), presents significant difficulties. In-person surveys can be costly and often miss the most vulnerable populations, including refugees, people living in informal settlements, and those working in the cash economy. Moreover, fear of adverse consequences, such as deportation for undocumented migrants, can deter participation. This lack of data makes it exceptionally difficult to assist those most in need.
The Role of Citizen Science in Poverty Research
Could AI be the solution? Combining AI with data collected by citizen scientists, also known as community scientists, offers a promising solution. The Togo example demonstrates the power of AI by combining geographical information with individual data from mobile phones. Researchers are now recognizing the untapped potential of citizen-collected data.
Thanks to smartphones, Wi-Fi, and 4G, an increasing number of people in cities and towns are actively collecting, storing, and reviewing their own social and environmental data. For instance, volunteer researchers in Ghana are gathering data on marine litter along the coastline, integrating this information with their country’s official environmental statistics.
Partnership and Potential of Citizen Science
In a recent article published in Nature Sustainability, a group of data scientists proposed that policymakers could utilize citizen-collected data alongside AI tools. The authors advocate for cooperation between AI researchers and citizen scientists.
International organizations like the United Nations Statistical Commission are now encouraging citizen scientists to contribute data, particularly for the UN Sustainable Development Goals (SDGs). The UN recognizes citizen science as a possible solution for the poor representation of hard-to-reach populations in SDG progress reports. Funding is critical to supporting citizen data collection efforts and expanding them with AI tools.
However, the United States, which is the largest national funder of data and statistics in LMICs, is withdrawing from several international commitments, including the World Health Organization and freezing foreign aid. While funding for official statistics stabilized after the pandemic, future support is uncertain if the United States pulls back.
Integrating AI with citizen data has various advantages. It empowers communities by allowing them to take ownership of their own information. Such accurate and well-curated citizen statistics may also enhance the quality of AI tools, working to remove the bias found in the training data. Also, AI has the potential to expedite the analysis of the data.
Conclusion
The safe usage of AI is absolutely essential, especially when assisting at-risk individuals or those living in poverty. AI should improve their lives without exposing them to new or worsening harms. Citizen-science data may be an important tool for providing the data necessary to make a difference. It is essential to support all participants in this research and to appropriately fund the research itself.
