Latest / World Food Prize Podcast: Dialogue That Feeds the World / World Food Prize Podcast with Rikin Gandhi of Digital Green
Transcript
- 0:11SPEAKER_01: You're listening to the World Food Prize Podcast, dialogue that feeds the world.
- 0:16SPEAKER_01: Twice a month, we bring you inspiring conversations with the leaders, innovators, and change makers working to increase food and nutrition security.
- 0:26SPEAKER_01: From groundbreaking research to global initiatives, we explore the ideas and actions driving food and nutrition security, sustainability, and opportunity for all.
- 0:37SPEAKER_01: Join us as we uncover solutions, share stories, and work toward a more food and nutrition secure future.
- 0:46SPEAKER_01: The Wellmark Foundation is proud to present the World Food Prize Podcast as part of our commitment to improving the health of all Iowans and South Dakotans.
- 0:55SPEAKER_01: Access to nutritious food is essential for living a long, healthy life.
- 0:59SPEAKER_01: And this partnership underscores the importance of advancing a food and nutrition secure future for all.
- 1:06SPEAKER_00: Hello, this is Michelle Hustain, president of the World Food Prize Foundation.
- 1:12SPEAKER_00: And before we get started with today's podcast, I want to invite you to something very special and meaningful to our foundation.
- 1:21SPEAKER_00: On Friday, May 15th, we're celebrating the 15th anniversary of our global headquarters, our home, the Norman E.
- 1:29SPEAKER_00: Borlog Hall of Laureates in downtown Des Moines.
- 1:33SPEAKER_00: The evening is much more than a gala.
- 1:35SPEAKER_00: It's a chance to help preserve one of Iowa's most inspiring and iconic historic landmarks.
- 1:43SPEAKER_00: And to support our mission of advancing global food security, we'd love for you to join us for our exciting casino night called High Stakes and Higher Purpose.
- 1:54SPEAKER_00: You can learn more about this fun evening and get tickets at WorldFoodprize.org slash Gala.
- 2:02SPEAKER_00: The World Food Prize team and I hope to see you there.
- 2:05SPEAKER_00: Thank you for celebrating with us and enjoy the podcast.
- 2:10SPEAKER_01: Please welcome Rickon Gandhi and your host, World Food Prize Foundation CEO, Thomas Vilsack.
- 2:18SPEAKER_02: Rickon, thank you so much for joining the podcast uh today.
- 2:23SPEAKER_02: Um let's start with Digital Green.
- 2:28SPEAKER_02: What is it and why did you name it Digital Green?
- 2:32SPEAKER_03: Digital Green started 20 years ago, initially as a part of Microsoft research in Bangalore, India, where we were trying to see what role could technology play for small techs.
- 2:45SPEAKER_03: And the name Digital Green came from our very first nonprofit partner was an organization called the Green Foundation, working with small farming communities just outside of Bangalore that were involved in biodiversity conservation and sustainable agricultural practices.
- 3:04SPEAKER_03: And like a lot of NGOs, they were stretched with human resources, finance, and they were looking for ways that technology could increase their scale and impact.
- 3:14SPEAKER_03: And so we partnered with them as our first NGO partner, and that's how we got digital green of adding the digital uh to Green Foundation's incredible work in the field.
- 3:27SPEAKER_02: Well now you're not telling me that twenty years ago you were involved in this.
- 3:31SPEAKER_02: That is true.
- 3:32SPEAKER_02: Were you about nine years old when you started this?
- 3:36SPEAKER_02: So good.
- 3:38SPEAKER_02: Were you working for Microsoft then or how did you get involved?
- 3:42SPEAKER_03: Growing up, I was inspired by astronauts.
- 3:45SPEAKER_03: Uh they brought together their unique combination of brains and brawn, and I sought to put together my own astronaut playbook by studying computer science, aerospace engineering, getting a pilot license, and about to enlist in the US Air Force when I had some friends who are starting up a biodiesel venture in rural India.
- 4:06SPEAKER_03: And that's what brought me to India while I was in the application process with these astronaut dreams.
- 4:13SPEAKER_03: And this biodiesel venture flopped within the first six months of me being there.
- 4:18SPEAKER_03: But I got to see how there's a small group of farmers who saw agriculture as a source of prosperity for themselves and for their families, while most saw it as a vocation of last resort.
- 4:29SPEAKER_03: And I then connected with Microsoft Research in Bangalore to see how we could investigate this further.
- 4:35SPEAKER_03: And what turned into a research project became an organization called Digital Green.
- 4:40SPEAKER_02: Well now, does your family have connections with farming?
- 4:44SPEAKER_03: We don't.
- 4:47SPEAKER_03: My parents, of course, did grow up in India, but uh don't have agricultural roots uh directly.
- 4:54SPEAKER_03: Uh but for me, I was just so impressed by seeing these heroes in their own right, a combining brains and brawn, just like the astronauts that inspired me to study science and engineering.
- 5:07SPEAKER_03: These farmers in India, they might be earning uh just uh two to three dollars a day, but in aggregate, they're producing the second highest farm output in the world.
- 5:16SPEAKER_03: And I it just seemed like uh a space and a community uh that I wanted to be a part of.
- 5:25SPEAKER_02: So Digital Green today is a is a nonprofit uh entity, correct?
- 5:30SPEAKER_02: That is right.
- 5:31SPEAKER_02: And uh just explain to the folks uh generally what does it do and who does it impact and affect and who does it work with.
- 5:39SPEAKER_03: So Digital Green is from the very beginning, when we were even a research project at Microsoft Research, we had this uh objective of seeing how could we increase the speed and effectiveness of agricultural extension programs or the training of farmers on improved practices and technologies that could boost their productivity and their resilience.
- 5:59SPEAKER_03: So we started with videos by and for farmers to share best practices where the first questions farmers would often ask is, What's the name of the person in the video?
- 6:08SPEAKER_03: Which village is he or she from?
- 6:09SPEAKER_03: To say that if this farmer could do it, then maybe I can too.
- 6:13SPEAKER_03: Uh and we were able to scale this work to about eight million farmers across Africa and Asia.
- 6:20SPEAKER_03: And then COVID happened, and then these in-person video screenings could no longer take place because of the physical distancing rules.
- 6:28SPEAKER_03: Uh but at the same time, many more farmers reverse migrated from cities, especially in India, to villages and took their phones with them.
- 6:37SPEAKER_03: And we saw about a hundred million views to our YouTube channel, and it was around the same time that ChatGPT came out three years ago, and that's when we saw the potential for flipping the paradigm of agricultural extension and enabling farmers to use local language voice, text, and images to use an application we call Farmer Chat to triage and diagnose issues on their farms directly on at their fingertips.
- 7:05SPEAKER_02: So in the midst of uh the devastation and and destruction of COVID uh to uh globally, uh you all found the opportunity out of that circumstance.
- 7:18SPEAKER_02: And you've created this the this this uh farmer chat.
- 7:23SPEAKER_02: What does it do?
- 7:24SPEAKER_02: I mean, so if I'm a farmer in India or I'm a farmer in Kenya or I'm a farmer someplace around the world, what are you offering me?
- 7:33SPEAKER_02: What can you do for me?
- 7:34SPEAKER_03: So farmer chat is a free Android app that's available on the Google Play Store, and it works across 16 languages from Amharic in Ethiopia to Hausa in Nigeria, and where farmers can use local language voice notes and images to be able to diagnose issues that they're encountering on their fields and also to forward plan for what they might want to market post-harvest or what they might plan to grow in the upcoming season.
- 8:07SPEAKER_03: What's been really exciting is we were doing this work with videos buying for farmers, but here now farmers can ask for information on demand.
- 8:16SPEAKER_03: And so they're asking, on average, between five to ten questions per farmer per month in sync with their location and season, and we're learning about what their needs and interests are.
- 8:26SPEAKER_03: And this AI assistant is then able to help them kind of go through their own learning journeys and of application and uh aspiration that these farmers are looking to grow towards.
- 8:39SPEAKER_02: What's the database of information upon which Farmer Chat draws to be able to respond and answer these questions that farmers are asking?
- 8:47SPEAKER_03: It's a great question.
- 8:48SPEAKER_03: There's multiple sources.
- 8:50SPEAKER_03: We partner with the National Agriculture Research and Extension Systems of Ministries of Agriculture in multiple countries.
- 8:57SPEAKER_03: Uh, and then we also integrate various types of weather services, soil information to localize the information that farmers are looking for, because they're not looking for generic ChatGPT or Wikipedia types of answers.
- 9:12SPEAKER_03: They're looking for information that's highly contextual for their location and season.
- 9:17SPEAKER_03: We also run what's called reinforcement learning, such that as farmers are interacting with farmer chat, the system learns not just automatically, but with local agronomists and local veterinarians reviewing the AI-generated responses and making corrections and augmenting the information so that it can be better aligned to the farmers' interests and also towards larger agroecological types of goals.
- 9:46SPEAKER_02: Now, are these uh folks who are reviewing the data and reviewing the responses, are they volunteers?
- 9:52SPEAKER_02: Do they are are they connected to digital green?
- 9:55SPEAKER_02: Do they are they hired by digital who are they and where where do they where do they live, where do they work?
- 10:00SPEAKER_03: Yeah, so we do a part a combination of both partnering with these ministries of agriculture and the International Agriesearch Network of CGIR, as well as uh retired uh agri scientists who are paid to do this uh job of reviewing the question-answer pairs that are coming in from a subset of these farmers' queries, and then making these corrections and then augmenting the knowledge base uh to better improve the system.
- 10:30SPEAKER_02: So, in a sense, this is kind of a two-way conversation with the farmer.
- 10:33SPEAKER_02: It's not just the farmer asking a question and getting a response, it's taking that farmer's question and uh sort of adding it to the database, if you will, and refining and perfecting the database.
- 10:45SPEAKER_03: Totally.
- 10:45SPEAKER_03: And we've had about 10 million queries of various types of voice, image, text forms.
- 10:52SPEAKER_03: And that data itself is so rich with the insights of these very diverse communities of farmers and sh indicates the types of challenges they're encountering, the types of interests they have across various crop and livestock systems, that at one level helps to prioritize what are the improvements that need to happen in these AI systems from in from tailoring the vocabulary, vernacular.
- 11:18SPEAKER_03: They have their own uh colloquialisms to refer to these uh terms.
- 11:23SPEAKER_03: And similarly, uh these farmers are have are taking pictures, not from the latest smartphones, but from their own uh kind of phones that they have on their fields, and being able to tune the image models to better support their the ways in which they're gonna take pictures of various crop stress is very critical for the agricultural domain and for these communities.
- 11:45SPEAKER_03: So, one part is to tailor the AI systems as per farmers' actual ways in which they want to use the system, and the second is also to use the same data to also inform researchers and policymakers about the needs and interests that the communities have, even separate from the use of AI.
- 12:04SPEAKER_02: So, what changed uh between uh pre-COVID and sort of the the day after COVID?
- 12:12SPEAKER_02: What what was it was it chat GBT?
- 12:15SPEAKER_02: Is that what changed it basically, or or was there something else that was taking place in the technology world that enabled you to do this?
- 12:24SPEAKER_03: It was a combination of these communities getting much more online compared to when we first started Digital Green 20 years ago, and we were just working with the Green Foundation outside of Bangalore, which of course is now seen as the Silicon Valley of India.
- 12:40SPEAKER_03: But back then, farmers were still using wired payphones in their villages, and they were just beginning to install the first cell phone tower in this village just outside of Bangalore's city limits.
- 12:53SPEAKER_03: And now most communities at least have one uh smartphone per household often.
- 13:00SPEAKER_03: It may be a shared device across multiple household members, but we find that uh penetration has really expanded.
- 13:08SPEAKER_03: And we've also been able to partner with some of the mobile operators themselves, like Safaricom in Kenya and Ethiopia, Econet, Wireless in Zimbabwe, which has been critical to be able to zero rate the FarmerChap application so that farmers don't have to pay the recurring data costs to interact with the application.
- 13:30SPEAKER_03: It's free for them to use.
- 13:32SPEAKER_03: And so that combination of having devices, having connectivity, and then, as you said, these uh the ChatGBT moment with large language models being as powerful as we've all been able to interact with them was the huge unlock.
- 13:48SPEAKER_03: And what's been interesting is like there's so many more young people who are back in agriculture uh as a result of not always finding employment in the cities, and they're with their phones uh looking for solutions to find new enterprises that they can start up in the real places that they operate in.
- 14:05SPEAKER_02: I remember my first trip to Africa was with uh Ambassador Andy Young, and we were we went to South Africa and we went to Nigeria and talking to farmers about uh new technologies and seed technologies.
- 14:17SPEAKER_02: And I'll never forget talking to uh one of the farmers from Nigeria and explaining to him that if they used uh the seed and if they essentially uh planted corn and then beans, but not plant planting corn and beans at the same time, that they'd actually increase their yield significantly.
- 14:34SPEAKER_02: And I can remember to this day the the shaking of the head, the the lack of confidence in what I was telling him because he was farming the way his father, his grandfather, his grandfather, his grandfather had farmed.
- 14:47SPEAKER_02: Do you do you are are you faced with that?
- 14:50SPEAKER_02: Uh when you basically say, boy, do we have a deal for you?
- 14:54SPEAKER_02: Uh we can give you something on your phone that can tell you when to plant and how to plant and what diseases are you're dealing with.
- 15:00SPEAKER_02: Do you do you are you faced with some skepticism about this?
- 15:04SPEAKER_03: Totally.
- 15:04SPEAKER_03: This is a uh, of course, a long-standing challenge in agricultural extension.
- 15:09SPEAKER_03: And I think the root of it is uh a trust.
- 15:13SPEAKER_03: How do you establish uh trust with these communities uh to say that this information is of relevance to you?
- 15:21SPEAKER_03: We found that in our video work with these videos featuring local farmers that fellow farmers could relate with and say that if this farmer could do it, I can too.
- 15:30SPEAKER_03: And the same has extended even into the farmer chat application.
- 15:34SPEAKER_03: The accuracy, of course, of these AI systems continues to improve as these frontier technology companies increase the state of the art of answering examination of like PhD and mathematics types of tests.
- 15:51SPEAKER_03: But we need to tune these models specifically to the benchmarks that farmers actually are interested in.
- 15:58SPEAKER_03: They're not looking for precision in a generic sense, they're looking for actionability and trust of what what is the what is the specific recommendation that I can actually take application of on my farm.
- 16:12SPEAKER_03: And then what is the result that maybe a fellow farmer has also experienced that I can also learn from?
- 16:17SPEAKER_03: So in Farmer Chat, we also surface fellow farmers' testimonials and experiences and even questions that if you're my neighboring farmer and you're seeing a tomato blight uh outbreak, then I don't need to wait to ask that question.
- 16:32SPEAKER_03: I can just be prompted to say, maybe one of my neighbors is seeing this tomato blight.
- 16:36SPEAKER_03: Would you like to learn more about it?
- 16:38SPEAKER_03: And and that kind of human connection remains super critical and is quite different from how we might interact with Chat GBT, which is we are just one-on-one.
- 16:48SPEAKER_03: But for these communities, agriculture is such a cultural uh vocation and is uh communal in a lot of ways where people are speaking with with one another uh at the chai stand or at the wholesale market, and building some of those ways in which farmers already interact into the system is super critical.
- 17:10SPEAKER_02: So I want to dig down a little bit deeper into how farmers are using this.
- 17:14SPEAKER_02: You mentioned uh the fact that uh they could potentially diagnose uh uh some issues with reference to their crop.
- 17:20SPEAKER_02: Uh tell me a little bit more about that.
- 17:22SPEAKER_02: How does that work?
- 17:24SPEAKER_03: So there's multiple ways in which a farmer could diagnose a pest disease or stress kind of event.
- 17:30SPEAKER_03: We find that farmers often are just taking photographs of their fields just to know whether they're even on the right track.
- 17:38SPEAKER_03: And it might might be that they're not even seeing any particular issue.
- 17:41SPEAKER_03: They just were taking photographs week to week of their farm and then are submitting it into farmer chat to ask the AI assistant, does it could there be some uh issue here or am I doing well?
- 17:55SPEAKER_03: And then when they have a problem, uh sometimes farmers often struggle with articulating what is this particular disease or stress that the crop or livestock is experiencing.
- 18:07SPEAKER_03: But being able to just take a photo and combine that perhaps with a local language voice note uh to say that I'm seeing some type of white bug uh on my crop uh of chili allows the AI assistant to be able to identify what crop the farmer is referring to and uh be able to have walk the farmer through trying to diagnose.
- 18:32SPEAKER_03: It doesn't always happen just on the first prompt.
- 18:35SPEAKER_03: It might come back to the farmer to say, hey, can you take a little bit of a closer uh picture or maybe take a farther uh uh photo of your field to see what level of uh pasture disease outbreak are we seeing here?
- 18:50SPEAKER_03: And in so doing, then uh the issue can be identified and then a recommendation can be provided.
- 18:56SPEAKER_03: And then the farmer can also have further questions around saying, for instance, uh maybe I can't purchase a particular input, but maybe I want to find a locally uh homemade type of remedy that I can uh perhaps make.
- 19:10SPEAKER_03: Uh and and so this these interactions turn out to be that type of data and insights uh that we then feed into the reinforcement learning process to make sure that the recommendations that come out are accurate and locally uh relevant, and then make the corrections as needed uh for these communities.
- 19:30SPEAKER_03: Uh but the exciting thing about AI is that it's a learning system.
- 19:33SPEAKER_03: So it's not just a one and done kind of thing.
- 19:36SPEAKER_03: It it learns with every interaction and with every improvement or feedback that is provided to it.
- 19:43SPEAKER_02: Is there any uh in terms of the makeup of the folks who are benefiting from this, is there any distinction between large and small farming operations?
- 19:52SPEAKER_02: I mean, I know you're probably dealing with primarily small whole or farmers, but can a large, fairly successful operation in India or Africa, an African country could could it use this uh this farmer chat?
- 20:06SPEAKER_03: Totally.
- 20:06SPEAKER_03: It is an or as a free application on Android, so any size operation could use it.
- 20:12SPEAKER_03: We do find that it is most useful, especially those who are into mixed uh crop livestock systems, and where it can often be a challenge to expect an extension agent who provide very essential services of access to seeds, credit, veterinary services.
- 20:30SPEAKER_03: Uh, but what Farmer Chap provides is a first level of triage to be able to support the farmer across a combination of the types of crop and livestock uh systems that they're into.
- 20:42SPEAKER_03: And then when they need in-person support or they need a physical input or physical access to off-take uh opportunities, they still uh connect to local cooperatives or a local uh extension.
- 20:56SPEAKER_02: Well, at the World Food Prize, we're obviously uh deeply interested in a world where there's uh food and nutrition security for all.
- 21:03SPEAKER_02: Uh and obviously that that lends itself to uh anything that encourages greater production and more sustainable production.
- 21:11SPEAKER_02: So uh uh FarmerChat, uh, can you also get information about conservation practices?
- 21:17SPEAKER_02: Can you also learn a little bit about how you might be able to do a livestock operation uh more effectively and efficiently with reference to uh the greenhouse gas footprint uh of that operation?
- 21:28SPEAKER_02: Is that also information that you could potentially obtain from uh FarmerChat?
- 21:33SPEAKER_03: Yes, and we are, I would say that we are also looking for partners, uh partners who can help us on some of this content and knowledge uh curation to support uh these farmers' interests, and partners also on the who can leverage FarmerChat for perhaps the constituencies of farming communities that they might be working with.
- 21:52SPEAKER_03: To give an example, we partnered with UC Davis, which has uh part of a global methane hub initiative for feed ration balancing.
- 22:00SPEAKER_03: Formulations and we're plugging in that model into FarmerChat so that the it's not just an AI generated response.
- 22:07SPEAKER_03: We can plug in the ground truth uh research work that these uh universities uh have developed to say what is the optimum balance of feed to feed to your cattle that can boost milk yields while reducing methane emissions.
- 22:22SPEAKER_03: Um, and we've seen great success uh in takeup by farmers when these interactions are much more on demand, that they're not being told to learn about this or that.
- 22:34SPEAKER_03: The farmers themselves are coming to the service and inquiring about what to feed to their animals to boost their milk yields.
- 22:41SPEAKER_03: Then you see a 4x increase in the rate of application rates of these practices as a result.
- 22:48SPEAKER_02: Well, I know the Global Dairy Platform, which is the large dairy processing uh and very large dairy co-ops, are working in Africa, just recently got a fairly significant grant uh to work with uh farmers in a number of countries to try to lower their greenhouse gas footprints.
- 23:04SPEAKER_02: Maybe there's an opportunity there for uh for connection between uh you folks and and uh GDP.
- 23:10SPEAKER_03: Uh just we'd love it.
- 23:13SPEAKER_03: Yeah, dairy and livestock is a very major uh area of interest amongst these communities, of course.
- 23:19SPEAKER_02: Um would love to so you affiliate you you are you can be affiliated with uh a number of uh organizations that represent uh blocks or groups of farmers.
- 23:27SPEAKER_02: I mean, the equivalent of the Farm Bureau or the National Farmers Union in these countries, you you you you connect and partner with that uh organizations like that to get the word out about uh about uh farmer chat?
- 23:39SPEAKER_03: Yeah, it's through a combination of both digital uh marketing to these communities on social platforms that many of these farmers now are on, like Facebook and WhatsApp uh and and TikTok, but also through these on-the-ground trusted networks that the farmers have, as you said, ministries of agriculture, sometimes uh some of these cooperative networks that exist in some of these communities.
- 24:04SPEAKER_03: And that's how we've been able to scale to 1 million farmers today, primarily across India, Ethiopia, Kenya, Nigeria, Brazil, and Zimbabwe.
- 24:12SPEAKER_03: And we're looking to scale this to uh 4 million additional farmers over the next year with the objective of seeing how we can crowd in other uh local actors, um, researchers, service providers, market uh actors to this uh platform of farmer chat.
- 24:29SPEAKER_03: Because of course it's not just about the information.
- 24:31SPEAKER_03: Farmers are looking for ways that they can take action and actually see tangible value and increases in income and resilience ultimately uh for themselves and their families.
- 24:41SPEAKER_02: Well, well, part of uh income uh generation is obviously markets.
- 24:46SPEAKER_02: Um is this something that a farmer could uh you know, could say, you know, I'm thinking about planning X this year.
- 24:54SPEAKER_02: Uh is there a market?
- 24:55SPEAKER_02: Uh is that kind of information also available under uh Farmer Chat market information?
- 25:00SPEAKER_03: There is some market information.
- 25:02SPEAKER_03: I would say that it is variable in terms of its availability across all the geographies where we operate in.
- 25:08SPEAKER_03: Um, but this is an area that we'd love to partner with other organizations who might be collating uh input uh price information and off-take opportunities for these communities.
- 25:18SPEAKER_03: Of course, that is ultimately what these farmers are looking for, is for more money in their pockets.
- 25:23SPEAKER_03: But where possible, we do uh link also some of the government uh information about some of the support prices that exist in some countries so that farmers might realize that there is uh market support or government support uh for certain commodities or certain uh even subsidies that they might be able to avail.
- 25:43SPEAKER_03: Uh, but it is a big challenge and it's certainly a need.
- 25:46SPEAKER_03: So if there are other organizations that have this kind of information, we'd be very keen to integrate with that.
- 25:53SPEAKER_02: Um you mentioned uh you've got goals.
- 25:56SPEAKER_02: Uh it started, as you say, in India and it's expanded now.
- 26:02SPEAKER_02: It's in Africa, it's in South America.
- 26:04SPEAKER_02: Um in terms of the the uh goals, do you also have metrics in terms of whether farmers have seen increased production as a result of this or increased profit or a combination of better environmental uh benefits?
- 26:21SPEAKER_02: Are uh is there any way of tracking that in what you all do?
- 26:25SPEAKER_03: For sure.
- 26:26SPEAKER_03: And we have three randomized control trials taking place uh of this work to measure changes in productivity and in income across Ethiopia, Kenya, and in India in partnership with organizations like the International Food Policy Research Institute, IFPRI, uh the Bordeaux School of Economics, ID Insight, and others.
- 26:47SPEAKER_03: We also have been able to look at this query log of the 10 million questions that farmers have asked of FarmerChat so far, and have been able to see trends like sometimes farmers might be just asking questions of curiosity across a breadth of different topics.
- 27:04SPEAKER_03: Whereas in some cases, farmers are getting really deep to say, I'm I want to set up a poultry coop and I want to learn where do I get chicks and vaccinations and feed or uh where I can sell uh these uh chickens.
- 27:18SPEAKER_03: And we've been able to do some initial uh social return on investment calculations to see like for those types of practices that farmers are really digging into, what does that translate into the types of changes that these farmers are seeing?
- 27:32SPEAKER_03: And uh we found that it's about a 42 to 1 ratio of for a$1 investment,$42 worth of uh return that the farmers are getting back.
- 27:44SPEAKER_03: Uh and that's because these uh AI models keep getting cheaper.
- 27:48SPEAKER_03: And it's not just like the large language models that are often very resource-intensive and costly to use.
- 27:55SPEAKER_03: Uh, the small language models uh of AI can be better tuned for our agricultural domain, the vernacular that farmers speak, and they're way cheaper and faster to operate, and they have this kind of ROI uh for farmers.
- 28:11SPEAKER_02: Well, let me make sure I understand this.
- 28:12SPEAKER_02: The the the the your your the farmer chat itself is free.
- 28:18SPEAKER_03: That's right.
- 28:19SPEAKER_02: All right.
- 28:20SPEAKER_02: But the farmer has to pay for information that's obtained through Farmer Chat?
- 28:25SPEAKER_03: Is that no that the information is free, but we're basically like through our nonprofit philanthropically supported uh investments making in covering the cost, right, of the AI solution to be servicing their free usage of uh FarmerChat.
- 28:42SPEAKER_03: So for a$1 philanthropic investment, the farmer is earning$42 uh on their farm for the types of practices that they're not the farmer's dollar that that's generating its it's it's your dollar or your donor's dollar.
- 28:56SPEAKER_02: Uh and that's a metric that you can give in raising resources to basically say to philanthropic organizations this is a pretty good investment because uh for every dollar you invest, farmers are benefiting$42.
- 29:09SPEAKER_02: That impacts not just farmer income, it impacts poverty levels, it impacts uh educational levels for children.
- 29:16SPEAKER_02: I mean, there's there there are a lot of rippling impacts and effects of uh farmers doing well.
- 29:21SPEAKER_03: Uh, and yeah, rural economy, yeah, rural economies in general.
- 29:24SPEAKER_03: And the cost of these AI systems is really dropping very quickly.
- 29:29SPEAKER_03: It cost us about 20 cents to serve a farmer per month a year ago.
- 29:35SPEAKER_03: Now it's about 20 cents per year, uh, as the cost has dropped less than like 15x uh over the last uh year itself.
- 29:43SPEAKER_02: So uh we only have just a minute or two left.
- 29:47SPEAKER_02: I want to take you uh from outer space where you were uh a number of years ago down here to Earth.
- 29:54SPEAKER_02: Uh is this, did you have any sense when you started on this trek 20 years ago that this is where you would be 20 years hence?
- 30:04SPEAKER_02: And how does it feel uh to see the progress that's been made in that period of time?
- 30:09SPEAKER_03: Certainly I would not have expected it.
- 30:11SPEAKER_03: Uh I think when I was growing up, I was looking at these biographies of astronauts and always was trying to piece together how I could sort of follow in their footsteps.
- 30:22SPEAKER_03: But in some of these biographies, what for these farmers who see the earth from above, they have what's called the overview effect and wonder and have questions about why is there war, why is there poverty here on Earth when it's you know such a special place in the rest of the blackness of space.
- 30:40SPEAKER_03: Uh, and some of them come back to the earth to reconnect with the world and its people as farmers, as teachers, and others.
- 30:48SPEAKER_03: Uh, so I didn't get all the way up there, but I am so uh grateful for having had this opportunity to partner with such amazing people, including yourself, Governor Vilsack, and all those on the ground, including these farming communities, uh, to be on this journey of learning uh that honestly we're still uh, even after 20 years, still in the early innings of uh and just trying to uh help enable these farming communities unleash their own agency and choice to become uh and what they aspire to be.
- 31:21SPEAKER_02: This has just been a fantastic uh quick half an hour of conversation about Digital Green and your tremendous work on behalf of smallholder farmers around the world.
- 31:31SPEAKER_02: So uh very much appreciate you being part of the podcast and uh uh anxious to see what the next 10, 20 years, if I'm still around, um, Digital Green will be accomplishing.
- 31:42SPEAKER_02: Tremendous story.
- 31:43SPEAKER_02: Thank you for joining us today.
- 31:45SPEAKER_03: Thank you for having me.
- 31:46SPEAKER_03: Great to be here.
- 31:47SPEAKER_01: Thank you for listening to the World Food Prize Podcast.
- 31:51SPEAKER_01: If today's conversation inspired you, we invite you to take the next step.
- 31:55SPEAKER_01: By making the gift, you help support dialogue like this, elevate innovative solutions, give voice to science, and invest in the next generation of leaders working to end hunger.
- 32:09SPEAKER_01: Visit WorldFood Prize.org to donate, learn more, and get involved.
- 32:14SPEAKER_01: Together through dialogue and action, we can sustainably feed the world.