AI and Citizen Science: A Powerful Combination in the Fight Against Poverty

Combine AI with citizen science to fight poverty

Artificial-intelligence tools and community science can help in places where data are scarce, so long as funding for data collection does not falter in the future. Of the myriad applications of artificial intelligence (AI), its use in humanitarian assistance is underappreciated. In 2020, during the COVID-19 pandemic, Togo's government used AI tools to identify tens of thousands of households that needed money to buy food, as Nature reports in a News Feature this week. Typically, potential recipients of such payments would be identified when they apply for welfare schemes, or through household surveys of income and expenditure. But such surveys were not possible during the pandemic, and the authorities needed to find alternative means to help those in need. Researchers used machine learning to comb through satellite imagery of low-income areas and combined that knowledge with data from mobile-phone networks to find eligible recipients, who then received a regular payment through their phones. Using AI tools in this way was a game-changer for the country. Now, with the pandemic over, researchers and policymakers are continuing to see how AI methods can be used in poverty alleviation. This needs comprehensive and accurate data on the state of poverty in households. For example, to be able to help individual families, authorities need to know about the quality of their housing, their children's diets, their education and whether families' basic health and medical needs are being met. This information is typically obtained from in-person surveys. However, researchers have seen a fall in response rates when collecting these data. Gathering survey-based data can be especially challenging in low- and middle-income countries (LMICs). In-person surveys are costly to do and often miss some of the most vulnerable, such as refugees, people living in informal housing or those who earn a living in the cash economy. Some people are reluctant to participate out of fear that there could be harmful consequences -- deportation in the case of undocumented migrants, for instance. But unless their needs are identified, it is difficult to help them. Could AI offer a solution? The short answer is, yes, although with caveats. The Togo example shows how AI-informed approaches helped communities by combining knowledge of geographical areas of need with more-individual data from mobile phones. It's a good example of how AI tools work well with granular, household-level data. Researchers are now homing in on a relatively untapped source for such information: data collected by citizen scientists, also known as community scientists. This idea deserves more attention and more funding. Thanks to technologies such as smartphones, Wi-Fi and 4G, there has been an explosion of people in cities, towns and villages collecting, storing and analysing their own social and environmental data. In Ghana, for example, volunteer researchers are collecting data on marine litter along the coastline and contributing this knowledge to their country's official statistics. Last December, a group of data scientists argued in a Perspective article in Nature Sustainability that these data could be used by policymakers in conjunction with AI tools (D. Fraisl et al. Nature Sustain. 8, 125-132; 2025). In the piece, Dilek Fraisl, of the International Institute for Applied Systems Analysis in Laxenburg, Austria, and colleagues call for a partnership between AI researchers and citizen scientists. The authors could be pushing at an open door. International organizations such as the United Nations Statistical Commission, which sets the standards for measuring official statistics, want more citizen scientists to contribute data, such as for the UN Sustainable Development Goals (SDGs), the world's plan to end poverty and achieve environmental sustainability. Hard-to-reach populations remain poorly represented in SDG progress reports, and the UN sees citizen science and citizen data as a potential solution. But making such close partnerships happen needs funding, both in supporting citizen-data collecting efforts, and in taking them to the next level with AI tools. This could be a challenge at a time when the United States, which is the largest national funder of data and statistics in LMICs, is withdrawing from international commitments, including exiting the World Health Organization and freezing foreign aid. Funding for official statistics started to stabilize after the pandemic, but the future will be less certain if the United States pulls back (see 'Data dollars'). Integrating AI with citizen data has many benefits. For one, it enables communities to take ownership of their information, knowing that it is their data that they are collecting and storing, and that the data will not be held by a third party. Accurate and well-curated citizen statistics could also improve the quality of AI tools, which often perpetuate bias or inaccuracies found in their training data. The use of AI also has the potential to speed up analysis of those data. AI has to be deployed in a way that maximizes benefits and mitigates or reduces risks. This is especially important when it comes to using AI that involves people who are vulnerable or living in poverty. AI has to make their lives better and not expose them to further or different harms. Citizen-science data might just be the medicine that the doctor ordered. All those who participate in this research must be encouraged and the research itself needs to be appropriately funded.

Can AI help beat poverty? Researchers test ways to aid the poorest people

Beginning in late 2020, with the COVID-19 pandemic in full swing, the mobile phones of tens of thousands of impoverished villagers in Togo pinged with good news. Their money was ready. With the help of artificial intelligence (AI), these villagers in the narrow strip of land in West Africa had been approved to receive around US$10 every two weeks, delivered directly to their mobile money accounts. Although that might not sound like much, it could keep people from going hungry. Conventional poverty-relief projects rely on data gathered through in-person surveys -- which were not possible during the pandemic. But Togo's effort, dubbed Novissi, which means 'solidarity' in the local Éwé language, incorporated AI to determine who should receive aid. The project, led by Cina Lawson, the Togo Minister of Digital Economy and Transformation, along with scientists from the University of California, Berkeley, and the non-governmental organization (NGO) GiveDirectly, analysed satellite images and data from mobile-phone networks to estimate the wealth of particular regions and individuals. "We needed a surgical approach," says Lawson. It was an important moment for the use of AI in anti-poverty work, she says. Some 700 million people around the globe live in extreme poverty, defined by the World Bank as living on less than $2.15 per day. Ending that poverty, one of the United Nations' Sustainable Development Goals, requires an understanding of who is in need and what their needs are. But measuring poverty has long been a challenge, in large part because of the time and cost involved in trying to collect data from the poorest and most vulnerable populations. AI allowed Lawson to leapfrog the conventional hurdles of using old and incomplete data to quickly make the most of her limited budget. It's an approach that is garnering both interest and controversy, says Joshua Blumenstock, a computer scientist at University of California, Berkeley, who collaborated on Novissi. AI tools can not only be fast, says Ariel BenYishay, a development economist at the AidData Research Lab at William & Mary, a university in Williamsburg, Virginia, but they can also include a larger, more representative portion of the population than household surveys do, and identify patterns in data that even specialists could miss. AI might also help researchers to evaluate how well programmes meet their objectives and demonstrate how investments in areas such as health, agriculture, education and infrastructure pay off -- or not. The World Bank recognizes this value and has been developing advanced AI tools to try to forecast food crises and violent conflicts, and to pull insights from large swathes of data gathered after an aid intervention. It concluded its Poverty, Prosperity, and Planet report in October 2024 by noting that anti-poverty "efforts should focus on leveraging machine learning and artificial intelligence models to close data gaps and enable more timely monitoring". But there are reasons to be cautious, says human geographer Ola Hall at Lund University in Sweden, who researches the intersection of AI and poverty. AI models have been criticized for being racist, sexist and otherwise biased. Just as household surveys often miss the poorest families because they do not have permanent housing, AI-driven programmes might not help individuals who do not have digital data trails, Hall says. They are nowhere near accurate enough to determine who qualifies for aid or cash subsidies and who doesn't, he says. However flawed AI might be, though, current systems of evaluating poverty are just as abysmal, says BenYishay. "The baseline isn't perfect data. It's actually very crappy data," he says. British social reformer Charles Booth undertook an early effort to quantify poverty from 1886 to 1903 when he criss-crossed London's cobblestones collecting data on people's incomes and social class. He created a colour-coded map of the city and reported his findings in a treatise titled Life and Labour of the People in London. English sociologist Seebohm Rowntree and his team interviewed 11,560 families in York, UK, and published the findings in a 1901 book called Poverty: A Study of Town Life. The team calculated poverty on the basis of the ability to meet a person's "physical efficiency", or their minimal nutritional requirements. A sample minimal diet might include bread, porridge, boiled bacon, potatoes, skimmed milk and little else. After US President Lyndon Johnson declared a 'War on Poverty' in 1964, the Office of Economic Opportunity adopted a poverty threshold devised by economist Mollie Orshansky that took a similar approach. It defined poverty as the bare minimum income required to cover food, shelter and other basic costs. Around the same time, India performed similar calculations for its populace. Although each expert tinkered with their formulae to account for local variations in rent and food costs, they all defined poverty on the basis of how much money a person lives on per day. The dollar-per-day approach is blunt and easy to communicate, says Dean Jolliffe, an economist at the World Bank in Washington DC. Yet, how much money a person spends to get by is just one aspect of poverty. Economist and Anglican priest Sabina Alkire advocates for a more nuanced way to define poverty. "I want to know how many poor people lack a house, how many poor people have a kid out of school, so I can actually respond in very tangible, direct ways," says Alkire, who is the director of the Oxford Poverty and Human Development Initiative at the University of Oxford, UK. In the early 2000s, Alkire wanted a way to capture poverty's various affects on people. Just because someone has enough money to buy food doesn't mean they have enough for medical care or school fees, says Alkire. In 2008, Alkire worked with James Foster, an economist at the George Washington University in Washington DC, to develop what's called the Multidimensional Poverty Index (MPI). The approach estimates a unified measure of poverty by tallying up the deprivations and their intensity, with a total of ten indicators, including nutrition, school attendance, access to drinking water and what a household uses for cooking fuel. For the field of poverty, this was a sea change. It allowed policymakers and others to measure, dissect and target the interacting variables that contribute to poverty at the household level. The United Nations Development Programme replaced its Human Poverty Index, which focused on survival, literacy and standard of living, with Alkire and Foster's MPI in 2010, although certain United Nations agencies along with the World Bank continue to rely on the dollar-per-day definition. Researchers and aid agencies have developed myriad ways other than the MPI to define poverty. These methods vary in the factors they include, depending on what they want to measure and the data at hand, says Jennifer Davis, who heads the Program on Water, Health & Development at Stanford University in California. In a 2024 paper, a team led by Davis and her graduate student Christine Pu evaluated four definitions of poverty that are used in the field, including daily per capita expenditure but not MPI, and found massive differences in how those definitions rank households in Ethiopia, Ghana and Uganda. "When we did our analysis, not only did we not find much agreement at all for the full sample, we didn't find it for urban households, or for the bottom 20%, or for the bottom 1% where we might expect the greatest need," says Pu. Along with the lack of agreement on definitions, there's the problem of time. Even a well-oiled field team needs several hours to survey a single family, says Jolliffe. Although poverty researchers have refined their metrics and are incorporating the latest computational methods for analysing data, they often continue to rely on on-the-ground surveys to collect those data. A lot of people are surprised that we still do household surveys, says Jolliffe. But, "this notion that we have data on everything about everybody is very much a rich-world perspective". As a PhD student in agricultural and resource economics, Marshall Burke was familiar with laborious data collection. To learn about farming and agriculture practices in East Africa, Burke travelled to Kenya and Uganda, where he spent months talking to farmers and walking their fields. But when Burke started the Environmental Change and Human Outcomes Lab at Stanford University in 2015, he wondered whether the computer revolution might offer better approaches. David Lobell, who had vast experience in remote sensing, occupied the office next to him. Around the same time, a specialist in AI and image recognition, Stefano Ermon, also joined Stanford University. The trio's discussions turned to how the ever-increasing data from satellite images could be used to help identify people living in poverty around the world. Knowing that night-time lighting can be a rough proxy for wealth, the researchers used night-time satellite imagery of areas across Africa alongside daytime imagery to teach computer models to identify features associated with wealth. Asking a computer to compare images of areas that are already known to be extremely rich or extremely poor is an electronic version of the game "spot the difference", Burke says. The algorithms compare the distribution and condition of roads, the amount of green space, the size and spacing of buildings and a multitude of other variables. "All the sorts of things you and I would think to look for in an image are a little bit predictive," says Burke. "A machine can sort through all that data," and determine what aspects are most relevant. In 2016, the team reported that the AI analyses of the satellite images correlated strongly with on-the-ground measurements of poverty. As machine learning advanced, Lobell, Burke and Ermon refined their models by incorporating the latest techniques. Using a pan-African data set of publicly available satellite images, the trio tested an updated approach in May 2020. When the team compared its machine-learning predictions with wealth-related survey data from 20,000 villages, the algorithm performed just as well as the laborious surveys, but at a fraction of the effort and cost (see 'Poverty predictions'). "This was a pretty seminal improvement concept for the development community," says BenYishay. Other teams are joining the experiment, throwing out a lot of different ideas, he says. Scientists are applying machine learning to look for patterns in satellite images and mobile-phone data, and to analyse the impacts of drought, agricultural productivity, infrastructure investments and more, he says. Much of this effort remains confined to academic research simply because the field is so young, says Abe Tarapani, chief executive of Atlas AI, a public benefit corporation in Palo Alto, California, co-founded by Burke, Lobell and Ermon. The stakes are too high to just dive in head first, says Tarapani. But when the pandemic hit, Lawson felt she had no other option. Lawson had a budget of $34 million for the Novissi project, from both government resources and NGOs, to help a population of more than eight million people. She needed to work out how to distribute it. But Togo didn't have poverty data that were recent or granular enough, especially for the country's agrarian regions. On a recommendation from an adviser, Lawson reached out to Blumenstock. In 2015, Blumenstock had reported using machine-learning algorithms to analyse mobile-phone data to predict wealth in Rwanda, and he'd been using satellite imagery to create poverty maps of all low- and middle-income countries at a resolution of 2.4 kilometres. Mobile phones are used widely in Togo, and the country has adopted mobile money. Blumenstock told Lawson that AI could distinguish between wealthier individuals and those living in poverty in rural areas on the basis of mobile-phone use, including mobile money transactions, as well as the frequency and diversity of calls and messages. This, combined with analysis of satellite imagery, provided a way to try to determine who most needed the money. An analysis of the Novissi project reported in 2022 showed that the AI approach better identified the people who needed aid than did other methods that were being considered in Togo. But, Blumenstock says, it's not yet clear how much better other approaches could have been with more resources and time to prepare. Other AI-powered efforts also face uncertainties. In 2022, a collaboration between the charitable organization Google.org and GiveDirectly organized money to be sent to the mobile money accounts of 6,000 families living in Africa. These regions had been deemed to be at high risk of future flooding by AI-powered forecasts. The money allowed farmers in Mozambique to fortify their homes and fields ahead of rising waters and, in a separate trial, post-flood aid was directed to Nigerian farmers in the weeks after flooding. But flooding in Mozambique didn't occur exactly where predicted, which meant that some unaffected families received money and some affected families didn't. These misses don't mean that the programme didn't work, says Daniel Quinn, a director at GiveDirectly in San Francisco, California. "The people just outside of the boundary line of what we were targeting were still deeply in need," Quinn says. Samuel Fraiberger, a data scientist at the World Bank's Development Impact Group in New York City, says that the effort held promise and was later expanded. "This is very much a real-world application of these kinds of methods," says Fraiberger, who is helping to lead AI efforts at the World Bank. Still, he is aware that without care in implementation, AI could perpetuate existing biases. Developing methods to evaluate the quality of data being used to train AI will be crucial, he says. Employing AI alongside big data also comes with privacy concerns; Lawson's team had to address such concerns while implementing Novissi. Alkire and others in the field remain sceptical of any efforts that might oversimplify data collection and analysis. "No country on Earth at the moment can do a multidimensional poverty measure from administrative records and satellite data, because you can't see inside the house and tell which kid is undernourished," says Alkire. And no amount of advanced computing can compensate for poor data, she adds. Hall notes that poverty researchers will always need a benchmark to check their algorithmic predictions against. Who needs aid, what form that need takes and what kind of assistance is appropriate is perpetually changing. Although algorithms are getting better, he currently rates AI's ability to identify individuals who are living in poverty at a "maturity level" of two out of ten. But Burke says both algorithmic approaches and conventional surveys are valuable and needed; he doesn't want to pit one against the other. When governments and aid agencies are facing limited budgets and sudden economic shocks, he says, AI could be the key tool that helps to get aid to the people who need it most.