Andrew Sweet, VP of Innovation at The Rockefeller Foundation + New Show Announcement: Intelligence for GOOD

Speaker A: Hi, everybody. This is Grant at, uh, Cause Artist. Welcome to Intelligence for Good. Over the last decade, through Cause Artist and the Disruptors for Good and Investing in Impact Podcasts, I've had the opportunity to interview hundreds of founders, investors, nonprofit leaders, and innovators working to solve some of the world's biggest challenges. But over the past few years, one topic has rapidly moved from the sidelines to the center of almost every conversation. Artificial intelligence. AI is already transforming how we work, learn, build companies, access healthcare, educate students, and solve global problems. Yet for many people, the conversation has become dominated by fear, confusion, and speculation. We hear about jobs being replaced, existential risks, and massive technological disruption. What we hear far less about are the people using AI right now to improve lives, strengthen communities, expand opportunity, and tackle some of humanity's most pressing challenges. That's why I created Intelligence for Good. This podcast explores the intersection of artificial intelligence, innovation, philanthropy, entrepreneurship, and social impact. Each episode features leaders who are building, funding, researching, and deploying AI in ways that create meaningful benefits for people and communities around the world. For our inaugural episode, I'm honored to welcome Andrew Sweet, Vice President of Innovation at the Rockefeller Foundation. Andrew leads the foundation's AI Partnerships and Strategy portfolio and sits at the intersection of technology, philanthropy, government, and global development. Before joining Rockefeller, he served as a presidential appointee at UH USAID during the Obama administration, worked with Dahlberg Advisors across Africa and the United States, and began his career as a Peace Corps volunteer in Togo, West Africa. At the Rockefeller Foundation, Andrew is helping shape how AI can be deployed to address challenges in agriculture, healthcare, education, workforce development, and public services. What's particularly fascinating is that the Rockefeller foundation has a unique connection to the history of artificial intelligence itself. In 1955, the foundation funded the Dartmouth Summer Research Project on Artificial Intelligence, the conference where the term artificial intelligence was first coined and where many of the field's earliest pioneers gathered to imagine what was possible. Seventy years later, Rockefeller finds itself once again helping shape the future of AI, this time focused on ensuring the technology serves humanity and expands opportunity rather than deepening inequality. In this conversation, Andrew and I discuss AI for smallholder farmers, the future of education, AI literacy, entrepreneurship, public services, global development, and why some of the most exciting applications of AI are happening far beyond Silicon Valley. Let's dive in. Awesome. Andrew. Well, thank you so much for joining me on this, uh, inaugural episode of Intelligence for Good is what I'm gonna call it, and what I'm gonna go with. I've interviewed, you know, thousands of people over the years across many different sectors, and all Of a sudden, AI is upon us. And there's a lot of discourse, uh, you know, good, bad, but also maybe a lot of confusion around a lot of things. So I want to start with, with. With your journey a little bit before the Rockefeller Foundation. Everything you guys are doing. And ironically, you guys are sort of at the birth of it 75 years ago.

Speaker B: I'll start this briefly with a career, I think. You know, I. I've always been inspired by big problems and, you know, always wanted to have experience outside of my own country. And so the first thing. The first thing I did was join the Peace Corps. So, uh, I was a Peace Corps volunteer in Togo, West Africa, as an ag extension agent and with zero AG experience. And so. But I was popped into a village with like 200 people and a bike and a stove and said, you know, this is your hut. See it, uh, two and a half years later. And so those types of experience, that experience has always been the most important experience I've ever had. And, you know, as with a lot of Peace Corps volunteers, you know, we're. We're drawn to public service. And so, you know, I spent some time at a think tank thinking through the future of US Global development policy. Always with kind of my experience in Togo. And. And then I had a great opportunity to join USAID as a political appointee during the Obama administration. And so, so I joined the Obama administration to, to first work on conflict issues in West Africa. Uh, and then eventually the became the senior advisor to the head of usaid, Raj Shah. So for a few years, I spent, you know, going everywhere with him to congressional testimony to Djibouti, Lebanon. Like every Congo, we. Every, Every week we were in a different place. Uh, and so that was an incredible opportunity and a very different opportunity from the Peace Corps, where I got to understand how large bureaucracies run, but also meeting people at. You never imagined that would meet. And just really being grateful for those opportunities, moved, uh, to South Africa to do some consulting at a company called Dahlberg and led the South Africa office for about five years. And then when the pandemic started, Raj Shah, my boss, was now the president of the Rockefeller Foundation. And so Raj said, hey, I want you to come to Rockefeller and to lead the pandemic response. And when the pandemic was dying down, we sat down and we said, okay, what's next? And in part because we have this great symbiotic relationship, in part because, um, we both like solving big problems, but mainly because of geography that I'm actually based in. San Francisco. We said, hey, why don't we double down on what we're doing in AI and uh, really think through how we can be effective. And that was just uh, immediately after the ChatGPT moment in November of 2022. And so part of the great thing about working at the Rockefeller foundation is, is we have an incredible history over the past 115 or so years. There's been a lot of moments that have kind of bent where the foundation, its employees have been ahead of the times. One of them that's less known is this moment in 1955 when we received a proposal from an academic at Dartmouth named John McCarthy. And in the proposal John McCarthy coined the term artificial intelligence. So the term artificial intelligence was coined in a proposal to the Rockefeller foundation in 1955. He asked for $12,000. And we said, you know what John? This idea is pretty out there about getting computers to think at the level or beyond the level of humans. We think we like your boldness and vision, but we're going to cut it to 7500. And so in 1956 we supported the, I think it was a 10 week Dartmouth summer conference in artificial intelligence and that included all of the luminaries in computer science at the time. So Marvin Minsky, who's kind of the uh, one of the godfathers of information theory, Claude Shannon, which anthropic named Claude after, was there and John McCarthy. So this year is the 20th or the, this year is the 70th anniversary of that conference. And on the 29th and 30th of October we're going to do a 70th anniversary conference with Dartmouth, bringing together today's AI luminaries to reflect on what has happened over these 70 years and how can we make sure that AI serves humanity and doesn't create further divisions within our societies.

Speaker A: Pretty incredible story. I want to talk about where we at now with the Rockefeller foundation. Maybe what's the mandate, you know, from your point of view, from your team's point of view, how do you look at the landscape this, this sort of commercial hit where everybody started to understand and being able to use it right from a commercial standpoint is powerful. What's your sort of mandate right now at the Rockefeller Foundation? What are you guys working on? What do you think is the most important thing right now to work on?

Speaker B: So my mandate, being the only Rockefeller person based uh, in San Francisco is primarily around relationship building. And so part of the mandate post ChatGPT moment was develop relationships with these companies, develop, build trust and Connect them to our partners solving problems around the world. And so, you know, being 115 years old isn't always seen as a good thing because it can be seen as, you know, you're stodgy, you're out of date. But the reality around this moment is that we've established an incredible set of relationships with governments, with nonprofit partners, with people that are intimate with the problems that we care most about. And so what I've, what I've found grant is that, yes, some of these nonprofit partners would like grant funding from us to have resources to solve problems, but a lot of what they want is access to this technology. They see the game changing potential of the technology and helping them to solve problems. If they're based in Nairobi or if they're based in Kampala or Bangalore, they may not necessarily have the relationships that my mandate is to build in San Francisco. So the initial calls that I got from our partners were not, hey, can I get a grant for this? It's, hey, do you know somebody here? Can you connect me here? So one of the best examples of this grant is one that was actually focused in the impact report. It's a partner of ours called Digital Green. And Digital Green's been around for about 20 years. It spun out of a project, uh, at Microsoft Research with the idea of how can we improve ag extension for farmers around the world. And this, as we were talking about earlier, is kind of near and dear to my heart as someone who dabbled in ag extension for my, for the first two years of my career with zero experience in rural Togo. And so what they said was, hey, we actually don't want funding from you. We would love, if we would like, we would love a connection to OpenAI. This was summer of 2023. So I said, yeah, sure, we can, we can do that. You know, we're agnostic with our partners. You know, I'm never going to say, like, you have to work with this partner. You have to. We're kind of responsive to our nonprofit partners. And so we made that connection. OpenAI deployed some engineers to build this chatbot called Farmer Chat. And In December of 2023, Sam Altman announced Farmer Chat at, uh, the first ever developer day for OpenAI. And it was, it just gets us, gives you a sense of how quickly things move, how relational things are. Fast forward to, you know, June 2026. Almost 2 million farmers are using PharmaChat across six countries in 16 different languages. And they have the goal of reaching 4 million by next year.

Speaker A: So quickly, for Those who, who aren't aware about PharmaChat. Exactly what does it do? Yeah, what is it for?

Speaker B: Farm. So it's an application where you can ask the app anything that's on your mind. You can take a picture of a crop and say, hey, what's wrong with this? You can say, hey, how can I make a pesticide that will help address this issue with the crop? It'll say, okay, you know, maybe I want to buy. What is the nearest store where I can go and buy this pesticide? It's, it's essentially an ag extension agent for every farmer in their pocket. And so part of the problem with ag extension as I witnessed firsthand, is there, there aren't enough ag extension agents around the world. Right. A lot of times the ratio is it can get to like high as like 1 to 5,000 in developing countries. And for communities that are fully dependent on agriculture for livelihood, that's kind of a dangerous ratio. So what we're saying is let's, let's have a personal ag extension agent in the pocket of every one armor so that any point in time when they have an issue, they can ask questions and get immediate responses. It's, it's just kind of uh, a, it's a transformational tool and it flips this narrative on the head ground a little bit about AI is going to take jobs away. What we're saying is what are the challenges where there aren't enough people to solve these problems, you know, in healthcare and community health workers, where there aren't enough, like how can we make sure that the technologies that exist are solving those problems where they're serious critical talent gaps and, and, and so agriculture is definitely, definitely one of those.

Speaker A: Part of why I wanted to do this show was because there's a lot of noise around job loss, around AI is sort of, you know, evil's obviously a strong word, but there's a lot of negative connotations towards the industry right now, and rightfully so. You could argue some of the leaders have not done a great job of, of presenting AI, uh, in a way that people can understand it, digest it and say, okay, how can this help me? How can this help our local economy, global economy, even to like you said, be, not take jobs but add maybe different jobs to the economy, enable billions of people to do things that they weren't able to do before that then in turns, you know, creates new jobs or whatever it might be.

Speaker B: And I would say grant, the uh, the, I would say the dynamic is quite different in developing countries. As it is from our own country. And so, for example, you know, every year there's a global AI summit. You know, this year the summit was the first ever global AI impact summit in India. And it, it, the, the dynamic in that conversation was very different. It was, it was the quote that I liked that stuck with me the most is we're not concerned about misuse. We're concerned about misuse. How can, how can I make sure that we're. This, this wave of technology is benefiting us and we are proactively identifying those solutions rather than think through, you know, um, some of the more tangential concerns. And so I, that, that really resonated with me. Um, and I think there is, you know, we talk, this is a little bit generic of a term, but we do talk about leapfrogging. And I think in many cases the opportunity that AI provides in developing countries to be more democratic, to reach more people, to flip that narrative, I think is, it's quite compelling. There's more energy, there's more optimism around the application of artificial intelligence in those contexts.

Speaker A: Do you see when you talk to whether it's global policymakers or governments or you know, nonprofit founders or for profit founders globally, do they have a hard time. Is there any disconnect locally from their countries with the, the models that are, let's say just OpenAI anthropo anthropic US Based. Is there any. Is like let's go India for example. Do they want to use local models to maybe develop certain things? I guess how does that. It can be. It's starting to be a discussion more and more about isolated models. And if we're only using two globally, that could be an issue. Right. But is do you see that there maybe India or any other country are saying, hey, let's invest in our local developers, build a local model to you know, to, to do some of these projects.

Speaker B: I think there's the term that you hear most around this topic is AI sovereignty. How can you make sure that the data sits within the country, that the models ah, are owned, that the, the languages within that country are represented in the model. And so I would say the, you know, the, obviously China is doing a lot of work with this. We saw how, how big the, the release of Deep Seq was. You know, I think it was last January when that happened. But in India's, India is a country that is, it's very confident in not only its capabilities, but its vision for the future. Right. If you look at a lot of even the major American tech Companies, they're run by, run by aliens. The, the university ecosystem in India is, it's incredible. The, the IITs. So you, they're, they're pumping out these engineers and I think frankly a lot of them are deciding to stay in India right now. And so you have competitors to the large American companies that are emerging in India and that, that uh, are focused on adding languages, making it more inclusive, making it more democratic. And I think that that will continue to, to be a key theme not only in China and India, but around the world. You know, one of the things that we did early on was we had a, we had a big convening in the fall of 2024 where we brought together all our nonprofit partners and our um, kind of Technology Partners, OpenAI, Anthropic, Nvidia. As a result of that, I went to Nvidia and I had some meetings over there, talked about the need for AI sovereignty in Africa. Out of those conversations, there was a big partner called Cassava Technologies that said, hey, we're going to buy these, we're going to buy these H2 hundreds. We're going to buy these GPUs and create the first AI factory for Africa. So now in, um, Cape Town, there's a large Data center with 2,000 GPUs. That is the purpose is to serve Africa. Right. And that's the opening sale. There's, they have plans to expand to I think, five more countries. But a lot of the countries now are saying, Graham, you know, we want to own our AI, uh, future. We want the sensitive data that exists in our country on financial flows, on health, proprietary information. We don't want that information to leave. And we want AI to help improve how we manage our systems. So this AI sovereignty piece is becoming quite a big issue. And I think a big transformation has been the establishment of the first AI factory in Africa and Cape Town, which became operational a couple months ago.

Speaker A: Wow. Uh, and a lot of both OpenAI and Anthropic. I've, ah, invested quite a bit of resources into grants. And I don't know if that's, if that's credits. Right. When you, when you introduce your, you know, partners or make introductions, are they asking for just, can you grant us certain token size or is it more? Hey, can you help us build this? Or is it, hey, can you. We're going to build this. We just need, you know, we would like to grant us tokens. Uh, I guess how's that, that impacts sort of grant making happening within this token ecosystem. Let's call It.

Speaker B: Well, I think there's a, there's a couple things. One, one is sometimes our partners would tell us that the relationships that we facilitate are more consequential than the grant money. And so we have a partnership in Rwanda, for example, with a group called Rising Academies. And they're using AI to help teachers with lesson plan. Their lesson plans are helping teachers to more effectively grade papers, but also serving as a tutor for students. Alan saw that in April of last year and it's great. Like it's a, it's going very well. The government of Rwanda is very supportive of the work. And I went and visited with Anthropic last year and it was kind of funny. They're like Rockefeller, like, thank you for the grant money. But actually Anthropic, what you're doing with your credits. And what they do is they have like a weekly conversation so that the engineers on the Rising Academies team can ask any questions of the engineers on the Anthropic, uh, team. So having access to that best in class technical capability and being consistent, that has transformed the technology. And I think that's a great example of kind of, as Anthropic would put it, beneficial deployment. Similarly, we have a partnership with Governor Moore in Maryland on, um, benefits access. And so a number of Marylanders have lost their jobs in part due to what's happening at the federal government. A lot of those folks were employed in the federal government and no longer have the jobs. There's I think 1.3 million people that receive public benefits. And you know, I think part of what we're trying to do is to not diagnose about what may happen, but be prepared. So if there is, if AI does in fact lead to job loss in the United States, that the social safety net has been strengthened before that happens. And so we introduced AI to what they call Maryland benefits. And so the 900 people in Maryland are constantly responding to requests about benefits. You know, a lot of times that information is quickly out of date. So they can use that chatbot to better respond to citizens and be more, uh, rapid in their response. And then eventually by the end of the year, we're going to launch with Maryland Benefits, a public facing chatbot so that at any point in time, you know, any Marylander can go into that system, ask what the state of the benefits are put in their information, and there may be even an agent that actually submits the application on their behalf. So just making it more seamless, making, making benefits access and delivery more human centered as, as, as we all Go through these painful processes with our own public institutions.

Speaker A: It's such a, it's such a mind bender too because you know, let's say you have uh, AI maybe displacing some jobs. It will happen, no doubt, right about, I mean it, it just, this happens with every evolution in technology or even an industrial revolution. But at the same time that AI helps build better access to the benefits that people get for that it's earned employment, whether it's Social Security, whether it's your Medicare benefits turnout. A company just raised 40, 35, $40 million doing this. Basically making the benefits ecosystem accessible for like senior citizens. Right. Or just non native. A, uh, tech person that's you know, a little bit older, it makes it easier for them to understand. My grandma's 95 years old. She, she lives by herself in a community and you know, her life is dedicated to just like dealing with healthcare providers, you know, and they have a person that comes like once a month and helps them out. But look, she's got an iPhone. Like she's, she texts me and Grammy emojis and stuff like she's dated. And so I do see, I do see the impact of like AI can help enable, you know, older citizens and just people eating benefits to access it easier, help it even understand it more. At the same time, it sort of creates that unemployment at some level too. So it's, it's definitely the shift that's happening that's, that's odd and interesting. But I think in it, at the end of the day it will, it will have a net benefit effect.

Speaker B: Yeah. And I think that um, the other thing, so there's the benefits in making sure that people like your grandmother can, you know, as easily as possible access the information she needs to be affected. Because I think so many, so much stress in our system is focused on navigating these archaic complex systems. And how can AI help improve that? I think the other thing, it's almost the opposite of that but complementary is how can we make sure that young people have the skills and confidence to succeed in this AI era? Because, uh, I think we don't want to get in the prediction game about AI is going to do this to this group of people. I think predictions are typically fraught. And so what we are trying to do is to prepare for eventualities. And so for example, in West Virginia we have this really great partnership with Governor Morrissey, with an edtech group called Thunkable and with Marshall University. Graham, what we're trying to do there and what we are doing There is, we've identified a number of schools in coal country where the governor said, well, we need to rethink the future of skills for these people. And we, and we need to do it early. So in, at high school level we've started a program with these, initially five schools in five counties, but uh, planning to roll it out to all 55 counties in West Virginia by the end of the year where Marshall University, who is led by a guy named Brad Smith, who used to be the CEO of Intuit. So it was very tech forward, said, hey, we need to bring these skills to West Virginians. So what they have done is they created a credentialing program. So these students in these five schools go through a 15 week program where they learn to code. So anybody can do this now through vibe coding and all these amazing tools where they learn to build applications to solve problems in their own community that they're facing every day. So not only is that a benefit, but there's also this thing where at after they go through this 15 week program, they get a micro credential to Marshall University at high school.

Speaker A: Love it.

Speaker B: You can go to Marshall University with a half a college credit already and confidence that you can do this or if you decide not to go to college, you can put this on your LinkedIn, you can put this on your resume that you have this credential from Marshall University, that you have this skill set. So part of it is rather than overwhelming and I think you got it right, Grant earlier, where sometimes the narrative from the tech leaders has been overwhelming about all this dramatic transformation. I just need to put food on the table. I don't even know what AI is. Is it like a glorified Google search? Like what is it? But so how can instead of being overwhelmed, create a sense of agency, create a sense of fluency? Like, okay, I think I can understand this technology. I see how it's proctored through my own life. I'm going to use it to solve my own problems, maybe create a business. I think there's a huge opportunity on entrepreneurship where you say, you know, uh, you know, I don't, I don't have an HR team, I don't have a finance background, but I can actually make a financial model with these tools like that. And so you can. What we're seeing is this kind of, this era of like, you know, solo entrepreneurs that are creating opportunity. There's a really interesting partnership that was announced with Anthropic in Workday and Lisk, where they're actually supporting these solo entrepreneurs to create new opportunities using artificial intelligence, knowing that you don't have to have all of these skills in order to do this, because you can get a lot of the capacity from artificial intelligence. My point being is it's good to do this earlier in one's career and it's good to inspire rather than overwhelm around artificial intelligence because it's not going away. You can't wish AI away. There was this thing that was signed by a bunch of leaders a few years ago where we need to stop the technology that Elon Musk and other signed that it ain't going to happen. So we have to deal with it. And how can we deal with it in a way that doesn't overwhelm Americans? How can we tell the stories about what it is, its applications, while being clear out about what some of the issues may be.

Speaker A: It kind of reminds me of uh, we're old enough to remember when, you know, the Microsoft Office suite came out and even the Adobe Creative Suite like we had. I had to learn all those tools. And I think these tools are very similar. Like if I remember filling out job applications, hey, are you proficient at Microsoft Office? And if you said no, then you probably wouldn't get the job, you know. And I think it's the AI literacy is going to be like that. Are you proficient in ChatGPT or Claude or whatever the, the platform might be for that specific job? I think the, the literacy part is super important. I think having this training in high school is, is on par. It should be stem. You need to add a STEM AI, right? You need to add that proficiency as early as possible because that is going to be what employers ask for. Are you proficient in this? If you're not, then I think it's going to be the same thing as Microsoft Office and all these other systems that came out, you know, two decades ago. And so I do think that it's great to see universities get involved, high schools get involved because I think as people learn early on then you have that spark of inspiration of hey, I could build stuff now. A stripe report I think a few months ago just came out and said Stripe Atlas, they're sort of the ability to, to create a uh, US LLC through their platform went up like 8x over the last year because. And they're all solo entrepreneurs, right? Because you, like you said, you can use these viacoding tools to build something you never thought you could build before. Build something that your local community needs even. It doesn't even have to be that big. It Just has to be something that people want and are looking for. And you can actually do that now, no matter what you're, you're proficient in or not proficient in, you can learn very, very quickly. And so that learning curve to me has decreased immensely, which is a very, very powerful thing.

Speaker B: Absolutely. And I think part. Part of what, part of what we have to do is to ensure that how can we make sure that AI doesn't become politicized in the way in which so many things in our side can easily become politicized. And so that's why I'm encouraged by these two examples like you have West Virginia and you have Maryland. You know, one is being led by a Republican governor, another is being led by a Democratic governor. How can you make sure that all the politicization of that's happening in a country that artificial intelligence doesn't get kind of caught into those camps? I think that's kind of, uh, you know, obviously we're just a, um, foundation in that space, but we're, we're helping with the narrative across the country to show how everyday Americans can solve everyday problems.

Speaker A: And what's also interesting parallel there too is that you have Maryland, very government heavy workforce. Traditionally, you would, you would not say maybe that workforce is as technically savvy as other industries. Likewise in West Virginia, uh, the mining workers and sort of, that they're sort of in this. In parallel, they're in the same position where the first round of technology sort of did not train them properly. There just wasn't systems in place. And not that they fell behind, it's just that now there is the ability to catch up very rapidly for those two industries. So that's kind of interesting too because they're, we're kind of in a similar spot.

Speaker B: I agree. And I think part of the new dynamic grant too is foundations like Rockefeller are critical in the sense that we have. I think we bring objectivity and credibility to some of this work. But what we're going to now see is this massive influx in philanthropic resources, in part due to the new entrance into the space. So there was a paper or a subset of that was out there a couple of weeks ago saying that there's going to be between 37 and $100 billion of new philanthropic capital every year that's going into this space. That's an incredible opportunity if it actually does flow to the problems that are most critical and that we care about. So how can we effectively deploy those resources to the problems that need it the most, you know, and so the Money is coming from the OpenAI foundation, it's coming from Anthropic. You know, I think all of the seven co founders of Anthropic have committed to getting away around um, 80% of their wealth. And they have one of the most ambitious matching programs in history for employees that want to do the same. So how can we make sure that those resources are actually deployed and they're going to things that actually impact everyday Americans? Because I think if we have this continued split of this technology is making this sub class of Americans incredibly wealthy and other Americans aren't benefiting from the technology that doesn't bode well for the, the, the social tissue of our society. I think they, there need to be some tangible, we have to, they have to say this is a breakthrough in health that occurred due to artificial intelligence, something that's real for your 95 year old grandma. Right. So it's not just, you know, this glorified Google search, it's actually applications to solve real human problems that otherwise wouldn't have been solved.

Speaker A: Yeah, the wealth divide could be a real problem in the next decade or two. But I think if done correctly we can actually see AI lessen the gap because I do think you can create wealth where it hasn't happened before for individuals who didn't have the opportunity or weren't capable before. I think that is the real opportunity is to lessen the wealth gap with AI. Uh, and that's a conversation that's not being talked about, it's only about the divide which obviously there's, there's going to be moments that are like that. But if we show that the possibility is there that unlike ever before, you can build something and change your life pretty rapidly. So I'm optimistic and I know that that's, that's not usually a place where people are within the AI industry. But I think everything that you have mentioned and talked about, I don't see being talked about enough around the opportunities that AI enables versus the opportunities that it kills. I'll end a little bit on this, on the future and obviously you're talking to some of the best companies, best people in the industry. What would you like to see over the next three to five years from the Rockefeller Foundation? What are some of the goals and successes you and the team would like to achieve?

Speaker B: Yeah, I think what we've, what we've, what we've started to do and what we've seen now is that we can have real impact with quantifiable metrics in our work. And so we Kind of know how to do this stuff now, right? I think, you know, four years in, I think there are some actual real impact numbers, including the impact numbers that I mentioned from Digital Green. Last month I was in Cali, Columbia with the mayor of Cali and with the local university. We created a tool called Dengue AI. And Cali, Columbia has been inundated with dengue and it's really ravaged them, particularly with the impact of climate change over in recent years where they, in our initial call, they kept on talking about rain bombs that were devastating the city. And after all these rain bombs, there'd be dengue outbreaks. So we partnered with the city and the university. Not big tech, by the way, like open anthropic. None of them were involved in this. It was all like data scientists within the mayor's office partnering with the local university. When I was with the mayor, we announced the results. They can now predict dengue outbreaks three weeks in advance with a 93% degree of confidence. And so that's happening in Kali. So part of what we're talking about is why is this only in Cali? Why isn't this across Colombia, Why isn't this across South America, why isn't this global? And so where those proof points exist, where those partnerships, those hard earned results, all the tensions that exist as you doing these things, all the tough decisions, we need to make sure that that knowledge gets out there and that we can scale these quickly. In part because it's just the right thing to do when you're actually solving a real problem problem, but in part because I think it lessens the anxiety a little bit around public opinion about what artificial intelligence may do to their livelihoods. So I think part of it is taking a step back and thinking about where the impact has happened and how can we make sure that we scale that impact as much as possible. So I think that's programmatically, that's a big piece. I think a second thing that needs to happen is on the narrative, as you and I were discussing earlier, there need to be more voices out there that aren't, that don't have a financial incentive to say one thing or another that can credibly and objectively say this is what's happening and these are some of the scenarios that we may face as a society. But right now there's a lot of noise and a lot of the noise is being dominated by people that benefit the most from the technology. And so how, who are the voices that need to be out there to tell Americans who tell people around the world and have an honest conversation with them. It's almost like the FDR fireside chats during World War II. You know, how like who, who is that person? Who are those people in the media environment is obviously very different than it was in the 40s. But I think we can draw inspiration from how he level set, how he humanized the war, how he rationalized the war to Americans who were very war weary for a number of years. I think that's a second piece. And a third piece is how, what are, how can we really transform how public services are provided? I think huge everywhere around the world and maybe most acutely in the United States. We're frustrated with the level of service, we're frustrated with the time, the complexity. It's almost like the system has been built to like to make us frustrated, to not work for us, to not work for us. And so how can we reimagine the public sector and how can we reimagine the benefits work that we had talked about? How can we imagine, reimagine just what the relationship between citizens and government look like? How can we redefine the social contract? How can we instill confidence that our government can still provide for our, uh, most vulnerable people? And I think we are in the game of optimism, rational optim optimism. And how can, what are those opportunities and how can we accelerate them and how can we communicate them? I'd uh, say those are kind of three areas that I'm personally excited about. And maybe the fourth one is how can we make sure that the new wealth that is being created can be deployed to help solve problems that are real and tangible for everyday Americans and not only go to increased inequalities between, between the world's first trillionaire and more people that are falling into the, you know, into poverty every, every year that that dynamic is not going to work. So how can we think creatively about deploying those 30, $70 billion each year in new resources so that people are optimistic about the future and not overwhelmed by the narrative?

Speaker A: Amazing. Andrew, thank you so much for taking the time. Amazing conversation.

Speaker B: Yeah, thanks so much.