Does artificial intelligence amount to a dystopian nightmare for the developing world, eliminating jobs and confirming the worse biases across societies? Or could it lead the way to a brighter future, helping to solve critical development issues?
We explore the ways in which artificial intelligence is being used across the globe at a grassroots level in India, Papua New Guinea – as well as get the big picture. Join The Development Podcast!
Tell us what you think of our podcast here >>>. We would love to hear from you!
Featured Voices
- Christine Zhenwei Qiang, Global Director, Digital Transformation, World Bank Group
- Naomi Longa, Seawomen of Melanesia, Papua New Guinea;
- Fred Munene, “Farm with Fred” farmer and YouTuber, Kenya;
- Snehal Joshi, Principal, Shikha Academy, India;
- Brigitte Hoyer Gosselink, Director of Product Impact at Google.org;
- Petra Molnar, Harvard Faculty Associate, lawyer, anthropologist, and author
Transcript
[00:00] Samuel Kwadwo Owusu Baafi: Hello and welcome to The Development Podcast from the World Bank Group coming to you from Washington DC. I'm Samuel. In this episode, the technology changing the way the world works. From farming to finance. AI, we explore the ways in which artificial intelligence is being used across the globe at the grassroots level. We'll hear from India, Papua New Guinea and Kenya.
Snehal Joshi: AI won't replace our teachers, but become a partner for them.
Samuel Kwadwo Owusu Baafi: But in addition to potentially unlocking economic growth, how do we ensure AI can work for all of us? The digital divide means a third of the world's population still doesn't have internet access. Will we see an AI divide? We get the view from the World Bank.
Christine Zhenwei Qiang: Maybe some jobs will be replaced, but hopefully more interesting and better paid jobs will be created.
Samuel Kwadwo Owusu Baafi: We also hear from google.org about their impact, partnerships and projects.
Brigitte Hoyer Gosselink: One of the things that I think is most important in getting these solutions right is that we're coming to those who best understand the problem, who are approximate to the communities that they're serving.
Samuel Kwadwo Owusu Baafi: And why AI can be a controversial tool when it comes to refugees. What will AI mean for development? Can it supercharge progress towards the sustainable development goals or will it deepen the existing divide? That's all coming up in this podcast.
Samuel Kwadwo Owusu Baafi: (Singing) This opera was not composed by a human being. This voice is not that of movie star Tom Cruise. (Seriously though everybody here very nice-) They were all generated by artificial intelligence. You'd be hard-pressed not to have noticed AI dominating the news headlines. While it has been around for some time, the technology has advanced at an astonishing rate in the last five years. It has also become available to the public at large through applications like ChatGPT, and the race to produce the most user-friendly AI tool is very much at the heart of the Silicon Valley tech titans to-do list. With it comes concern from many over the future of jobs with fears that AI will replace the human in some functions. There's also fears over impartiality, accuracy and the data that AI is trained on, not to mention the wellbeing of the humans that are manually labeling tens of millions of pictures to create the data sets. The carbon footprint of AI is also troubling many. As the use widens the data centers needed to house our digital information generates a lot of carbon emissions, and AI is increasing that footprint. Despite worries over how people will interact with AI, which we'll talk more about later, there's a good deal of optimism over how AI can revolutionize the way we work, provide low cost solutions to difficult problems, teach us new skills and boost economies, and help them diversify.
[02:59] Samuel Kwadwo Owusu Baafi: We may be some way off sitting back on the beach enjoying our leisure time while the machines do all the work, but AI can be a tool to help us do our jobs better. As our producer Sarah Treanor found out.
Sarah Jane Treanor: The enticing sound of tropical waves in Kimbe Bay, Papua New Guinea. The area is a biodiversity hotspot with many coral reefs and also the incredible marine life they support. It's the home of Naomi Longa, a conservationist passionate about protecting the reefs of Papua New Guinea and beyond.
Naomi Longa: We empower indigenous women and local women around Papua New Guinea with the training skills and resources that they need.
Sarah Jane Treanor: Her NGO, The Sea Women of Melanesia helps women to establish marine conservation areas where they can take the lead in managing their precious ocean resource.
Naomi Longa: And these are the areas where people or the local people already have that connection to the reef and they own the reef.
Sarah Jane Treanor: To Naomi respecting the deep cultural ties that these communities have to coral reefs is vital to her work, but also is artificial intelligence. Naomi is marrying traditional knowledge and practices with cutting edge technology.
Naomi Longa: Yeah, with the Reef Cloud AI software that we are using, we upload those images onto the project and then the AI crawls through all these images and speed up the data, letting us know how many crawl cover estimates it's got. If there's a disease or a bleaching happening, it will also spit out those data. We can have different data from different era, and then we can compare the changes happening. So if the crawl cover is decreases, then we know instantly from that software.
Sarah Jane Treanor: Most of the areas Naomi works in do not have good or even any data connections, but she still manages to use AI to gather vital information. She has to process it later on. Naomi is also using the technology to store and translate traditional songs, stories and culturally significant pieces of information alongside the reef monitoring and modeling data. This song is one from her village. It's called Tahe.
Naomi Longa: Yeah, very useful tool and now we are going to have these, what it's called the dashboard community dashboard, where we can get information from the community, not just general information about the coral reef, but also if they have a story or particular side of interest to them then we can also collect this information. It is also available for younger generation to utilize.
[05:50] Sarah Jane Treanor: Let's head over to India now where AI is being trialed in many different educational settings.
Snehal Joshi: Hello, I'm Snehal Joshi, principal of Shikha Academy in Mumbai, India. Shikha Academy caters exclusively to children from financially disadvantaged backgrounds. We have a research and development institute in our school called the Shikha Institute of Education and an AI tech development company called Shikha Labs. We have piloted an AI teacher coach named Saki at the school. We are testing Saki for personally coaching our teachers to make their pedagogy more student centered. We have also developed gen AI apps in-house, such as lesson plan assessors, discussion guide generators, rubric creators, and assessment generators. These tools help augment our teacher capacity and also save a significant amount of planning time and stress. Our team definitely finds it extremely useful. I think AI will surely play a significant role in India's education field. It can help us leapfrog Indian education by quickly scaling the best practices. However, it is essential to be mindful that while we can offload some tasks with precise instructions, mentoring and some higher order practices will still rest with humans. AI won't replace our teachers, but become a partner for them.
[07:36] Sarah Jane Treanor: And last, by no means least, I asked one young Kenyan farmer how he feels about AI and how he thinks it will encourage others to consider a profession in agriculture.
Fred Munene: My name is Fred Munene. I'm a young farmer and a social media influencer going by the name Farm with Fred. In my case, I'm using it to monitor the weather pattern. I'm also using it to monitor pest and any diseases in my farm. And in the near future I'm planning now to take the next step of adopting or integrating mechanization and AI in my farm. That is the use of drone, that is the use of AI driven tractor because the future is AI. When we talk of technology, we cannot miss AI. AI is a tool that will drive the technology that you are adopting or the technology that we plan to adopt, to make it easy and to be able to eliminate inefficiency. And I'm very happy that agriculture is taking the center point in adoption of technology to be able to produce more, to produce efficiently and to produce at a less costs.
[09:05] Samuel Kwadwo Owusu Baafi: Now, those are some really interesting snapshots of how AI is currently being used at the grassroots level and also gives us some idea of the versatility of the technology at hand. But what about the big picture? The IMF in January of 2024 estimated that AI will impact about 40% of jobs globally replacing some and complimenting others. So is this the new industrial revolution? I invited the World Bank's, Christine Zhenwei Qiang, to our studio in Washington DC to find out. We hear a lot about AI from the dystopian sci-fi thing, terminator and all that, to really enhancing impact on the scale of the industrial revolution, pretty much. So how is AI already being used in the developing world and how do you see it expanding across a range of areas where the World Bank is working, like education, healthcare and agriculture?
Christine Zhenwei Qiang: Artificial intelligence is nothing short of revolutionary in emerging and developing economies. If you think about the scale and also the speed that digital technologies reach to people, it took 75 years for fixed telephony. But for ChatGPT it was only two months. So it is really unprecedented. For example, in education, AI can bridge the deficits caused by teacher shortages. So AI can really help with customized class work plans and also enable self-guided learning programs that really help in improving education equity, but also quality. For example, in India, students using AI powered personalized learning tool scored higher in Hindi and in mathematics. Maybe one more example would be in the agriculture sector. We have seen a lot of the drones and the satellite. They enable collection and classification of information and images that allow farmers to oversee large pieces of land to manage soil health, optimize water, fertilizer, and pesticide levels. This way they can boost crop yields and also contribute to food security. I was having a chat with Minister Tijani of communications innovation and digital economy from Nigeria, and he actually mentioned to me that digital devices with AI enabled apps made farming jobs so much more attractive to young people in his country. So to me, this is so important because we need to generate a lot of jobs, especially in Africa for youth.
Samuel Kwadwo Owusu Baafi: I'm glad you mentioned Minister Tijani.
Christine Zhenwei Qiang: He's very inspiring.
[11:49] Samuel Kwadwo Owusu Baafi: Very inspiring. But we can't talk about AI without talking about risks. There are risks inherently in every technology and in AI it's particularly around cybersecurity, misinformation or privacy. Now, what are some of the measures that we can take to mitigate these risks, but above all, which is really interesting to me, the ethical considerations. What are the ethical considerations to keep in mind so that AI actually meets the needs of the developing world?
Christine Zhenwei Qiang: Yeah, it's very important that we put in careful guardrails to ensure that AI is applied ethically, equitably, but also inclusively. The first big concern is about misinformation and deep fake and the quality of data fed into the AI systems is crucial because AI models really rely on this data to learn and make prediction. Poor quality data can lead to distorted and incorrect predictions and outcomes. To prevent this, we are including validation layers that check the output against factual data and also assess consistency. There is also the issue of algorithmic biases. AI systems can sometimes perpetuate existing biases because they learn from historical data that may reflect societal prejudice. Just to take an example, if we are using AI model for recruitment, then we need to make sure that the data that we are using is really diverse, and also that covers all demographic groups. In addition, we should implement measures for algorithmic transparency, conduct regular reviews or audits to identify and correct any biases that may arise. AI relies heavily on data again, and which can lead to privacy breaches if not handled properly. So establishing robust data governance frameworks that prioritize user consent and data security is essential.
[13:58] Samuel Kwadwo Owusu Baafi: Let's talk jobs, labor market. That's what we're all scared of. I'll speak for myself. It terrifies me and as a communicator and they say we're safe, but you never know. So one of the really early concerns about AI is the impact that it's going to have on the labor market. Machines replacing jobs that hundreds of or maybe thousands of people do in one factory taken over by machines. Such jobs have been in abundance, especially in the developing world, just because there's so much cheap labor or affordable labor in those markets. How can countries and institutions like the World Bank confront this challenge?
Christine Zhenwei Qiang: There's such a big demand for creating jobs for youth, especially in Africa, in South Asia, in East Asia as well, and we all know that AI can augment workers in some tasks, but replace workers entirely in some other tasks. But overall impact, it's really hard to predict at this point, but when we look at the history, in the past the introduction of innovative technologies often actually expand job opportunities and give rise to novel and better paid occupation. More than 60% of employment in the US in 2018, those job titles they didn't exist in 1940. And also in Latin America, half of the 20 fastest growing skills are linked directly to new technologies. And in India, the demand for AI related skills has been growing so exponentially since 2016. So all of these statistics somehow give me hope that new technologies actually will expand opportunities. So I'm just trying to say that maybe some jobs will be replaced, but hopefully more interesting and better paid jobs will be created. So that's what every government need to focus on is to really skill up and re-skill workers so that they'll be ready when these new jobs become available.
[16:11] Samuel Kwadwo Owusu Baafi: Thank you again to Christine for coming to talk to us in studio. Let's hear a bit more about the kinds of programs and partnerships where AI is taking the lead. Sarah Treanor caught up with Brigitte Hoyer Gosselink, a director at Google.org.
Brigitte Hoyer Gosselink: We have a number of applications here. We work in partnership with NGOs. One of the things that I think is most important in getting these solutions right is that we're coming to those who best understand the problem, who are approximate to the communities that they're serving, and then we work to enable them with funding of course, but also with technical expertise to develop those solutions.
Sarah Jane Treanor: Give me an example of some of the partnerships Google.org has worked on that you've found particularly impactful or inspiring.
Brigitte Hoyer Gosselink: We've worked with organizations like Wadhwani AI in India who is using a tool to do pest identification and provide better information to agricultural extensionist and to farmers directly, but also to agricultural ministries across state and national government level to help understand what's happening in crops and allow for folks to better adapt the early studies and that show that not only were farmers able to increase their outputs and therefore their economic gains, but also to decrease pesticide use, which has of course the ancillary benefit on the environmental side. We've also seen big huge projects on the climate action front, projects like Climate TRACE, where we began, brought funding and our technical expertise to a set of organizations looking to really create real-time tracking for greenhouse gas emissions at a global scale, and to do that in a way that provides totally open data down to the site level. And then finally, really exciting work that's happening on the climate resilience and adaptation front. Obviously seeing an uptick in crises around the world, one of the most challenging areas where we've seen that is in flooding, which impacts just millions of people's livelihoods and lives. And Google itself has actually developed new AI models that provide new flood forecasts up to seven days in advance across 80 countries. We cut that information out on our platforms through tools like SOS alerts, but we also are working through Google.org in partnership with NGOs like the Red Cross, to really make sure that those alerts are provided all the way down to the community level. And then even doing innovative models like experimenting with, if we can do cash transfers in anticipation of a crisis, might people actually be able to move out of harm's way and therefore reduce the damage.
Sarah Jane Treanor: We heard earlier about a really innovative project in Papua New Guinea where data was being collected and then the AI was modeling the health of coral reefs. And in a lot of the communities working on this project, there was no capacity to be online at all. They took the data and uploaded it at regional offices, which was really time-consuming and involved physically traveling quite a long way by boat. So how do we make sure that the digital divide isn't widened even more with some parts of the world using tools that others simply can't?
Brigitte Hoyer Gosselink: Yeah, we will definitely need to see continued investment in infrastructure to really fully realize the potential of AI, but I am also really excited about some of the work that's happening to move AI models data collection closer to the edge. So it used to be we were with most things running these only in the cloud, only in big data centers, and that's still the case for a lot of the more advanced models, of course, that we're serving with things like Google's Gemini product. Google Gemini is our large language model enabled tool that allows you to access information and ask questions, but ultimately, actually a lot of this work is now able to be done on device. So actually when we launched Gemini, we launched a specific model that's meant to run on device. It's much smaller, it's slightly less capable, but it's actually still quite useful for many, many use cases. And in addition, the ability to store data, to have uploads to work kind of through the sneaker net, right, with passing around this information from disconnected places, ultimately maybe bringing it back to a hub that has that connectivity. We see organizations doing that all the time, and I think we need both things. We need to invest, of course in the infrastructure to build out that connectivity goal and we shouldn't wait until that happens. And so the combination, I think of the technology advancements with the innovative ways that people are using those, it just gives us a true advantage and today really seeing that AI opportunity come to life even while we're still trying to solve some of those infrastructure gaps.
[20:35] Sarah Jane Treanor: As we heard earlier from Christine at the World Bank, speaking to Samuel, there are many cautionary tales when it comes to AI being rolled out in various situations across sectors. One of those is the movement of people across borders. For years, there have been stories of AI failing to detect the facial features of people from various ethnic groups. There are also says Petra Molnar, human rights lawyer campaigner and author of the book, The Walls Have Eyes: Surviving Migration in the Age of Artificial Intelligence, bias issues to address and urgently. I caught up with her and asked her how AI is currently being used in the context of refugees.
Petra Molnar: AI and new technologies are now impacting virtually every single aspect of a person's migration journey. And different types of projects that you might interact with before you even cross a border are things like, for example, social media scraping of your activities online that different states are doing. There's different types of technologies that you might interact with should you find yourself in a refugee camp or a humanitarian emergency setting like biometrics. There's different surveillance and AI driven surveillance in particular that we've been documenting across different borders. And then even on the backend, so to speak, once you've already arrived in a country, you might be impacted by visa triaging algorithms that use machine learning or AI or voice printing technology or even AI lie detectors that we've been documenting where facial recognition and emotion recognition is used to try and determine whether you might be lying or not during the course of your interaction with a border official. The risks around people's human rights are very, very high. We know, for example, that facial recognition technology has trouble with people who are darker skinned or we know that there are biased algorithms that are being used because they also reflect the biases of our current world.
Sarah Jane Treanor: Petra, I read an article from you where you asked a generative AI system to show you pictures of a refugee and the results were very interesting. Can you explain to me what you got back and why it matters?
Petra Molnar: Yeah, sure. So I was trying to see how generative AI could possibly think about people on the move, migrants and refugees, and I asked it a couple prompts when I was finishing up my book, and I was curious to kind of see how different programs would respond to the question of what does a refugee look like? And it was really stark, but perhaps not surprising that the images that came back were either extremely racist depictions of people with black and brown skin in distress, or very kind of naive portrayals of smiling, vaguely Middle Eastern looking children just out there in the world. It was really quite depressing, frankly, to see that yet another kind of tech tool was relying on very, very discriminatory depictions of a very complex population.
Sarah Jane Treanor: You feel there needs to be regulation across the board when it to the application of technologies in relation to migration and refugees, but where do you feel there is a possibility of AI actually working well when it comes to the issues that we've discussed?
Petra Molnar: I definitely think it could, and this has been a challenge for someone like me who's spent years looking at all the human rights violations and the difficulties with using AI and new tools, but there's also another side to this story, and that is definitely thinking creatively about how technologies can be used to level the vast power differentials that are present in our world and empower mobile communities and give them access to resources. Last year, we started a fellowship program for people on the move to tell their own stories on the impacts of border tech, and honestly, it's been really interesting because last year, for example, half of the fellows were looking at surveillance and the difficult impacts of tech, but then the other half was looking at the more positive uses of technology. For example, the use of chatbots to make information available for refugees and mobile populations very quickly, or the use of technology for psychosocial support and digital archiving of community stories. There's this whole rethinking that we can do also when it comes to tech, but it must be led by, again, people on the move and affected communities that are the ones who really are the experts when it comes to having these conversations, because they're the ones who are being impacted by this every day. It's to me, ultimately, again, about highlighting the human impact and thinking about how technology could potentially be used to make people's lives easier and better, assist them in their asylum claims or even support them psychosocially. I think there's definitely openings for that.
Samuel Kwadwo Owusu Baafi: Well, this is not a topic that's going anywhere, and we'll come back to it, I'm sure. But in the meantime, if you have any thoughts on what you've heard, we'd love to hear them. Get in touch at The Development Podcast at worldbank.org if you'd like to send us an email with your take on the AI revolution. We promise you that a human being will read them. Thank you so much for listening, and we'll be back very soon.
ABOUT THE DEVELOPMENT PODCAST
This international development podcast brings together the data, research—and solutions—that can pave the way to a sustainable future. Through conversations focused on revealing the latest data, the best research, and cutting-edge solutions, let us introduce you to the folks working to make the world a better place. Don't miss an episode! Listen and subscribe for free on your favorite platform. And rate our show! ;)
Tell us what you think of our podcast here >>>. We would love to hear from you!
ABOUT THE WORLD BANK
The World Bank is one of the world’s largest sources of funding and knowledge for low-income countries. Its five institutions share a commitment to reducing poverty, increasing shared prosperity, and promoting sustainable development on a livable planet.
