Paul Ford: Hi, I’m Paul Ford, and usually I am joined by my co-founder, Richard Ziade. Rich is a little under the weather, and he was going to be here in the studio, and yelling at me as normal, and I said, “Could you please just get well?” And with much regret, he’s not here. So it’s just me—but not alone. Not alone at all. You’re listening to the Reqless podcast. I’m going to play the theme song, and when we come back, I’m going to introduce you to Perry Hewitt.

[intro music]

Paul: Okay, you’re here, Perry. Hello.

Perry Hewitt: Hi. Good to see you, Paul. And sorry to miss you, Rich.

Paul: Oh, I know, I know. He was sad, he was sad to miss you, too. He’s, he’s definitely one of the, like, [wheezing as if on death’s door] “I’ll come in. It’s fine!”

Perry: [laughing] Please don’t. Please don’t, Rich. Stay home.

Paul: Exactly. So Perry is someone who… I’ve known you for a while. Rich has known you for a while, too.

Perry: Yeah, probably 15 years.

Paul: Yes, exact— [noise of realization] Oh, wow. That just, that hurt a little bit.

Perry: Early Twitter days. Back when I was analyzing tweets at Crimson Hexagon.

Paul: I had smaller pores. [laughter] So anyway, I’m going to get—let me not screw up your title. Tell me what your job is.

Perry: My job is Chief Marketing and Product Officer at Data.org, which is a platform for partnerships to build and accelerate the field of data and AI for social impact.

Paul: That was, you know, it seems like you’ve rehearsed it. [laughter] You know what I love about this, too? It’s, like, marketing people, leaders in marketing and product, they get that script down. Like, you got to be good at that part. And then also because you’re on so many panels.

Perry: Repetition is key.

Paul: Yeah, and they just, you go down that panel. You got to get it out there. And also, the LinkedIns of your cohort are really, really good, almost always. They have to be, right?

Perry: Yep. Gotta stay on it.

Paul: Do you—how much time do you spend on LinkedIn?

Perry: Far too much time.

Paul: [whispering] I know, it’s just horrible.

Perry: Yeah, yeah. And I’m deeply ashamed of most of it.

Paul: No—

Perry: You know, it’s, yeah, it’s just…yeah, it’s a lot. It’s a lot.

Paul: Well, many, many nerds are going to now say, “Hey! You know, I heard you on Reqless,” and you’ll connect because that’s what you do.

Perry: Absolutely.

Paul: Okay, so Data.org. Let me try to explain Data.org in a fun, entertaining way, and you can tell me how I get it wrong.

Perry: Excellent.

Paul: Okay. Data.org is a not-so-big not-for-profit, but it gets support from sort of large organizations, and its focus is on helping people in the world make sense of data. So especially kind of do-gooder things like, you know, people who are gathering information about health outcomes in parts of Africa or India. And it says, hey, you’re getting all that data. That’s nice. A lot of times people will fund you to go get that data. And good for you. Now you have all that data. There are all kinds of practices, all kinds of things you can do with data. And we would like to help you use those best practices to get better outcomes. And we’re kind of the clearinghouse for that sort of thing. Is that at all correct?

Perry: I’ll give you a B minus on that.

Paul: Okay. [laughing]

Perry: So, you know, we do do data, we do do do-gooders, but what we really focus on is capacity development in data for social impact. So basically what that means—

Paul: [overlapping] Okay—

Perry: [overlapping] Let me, let me tell you in English. I know, I know—

Paul: [overlapping] Yeah, yeah. Civilians!

Perry: [overlapping] Sorry, I know, it was buzz-wordy.

Paul: [overlapping] Civilians listen to this show.

Perry: That was buzz-wordy. I take it all back. Capacity development means we want the people who are affected by algorithms, data software, to be in the driver’s seat, to be able to use the data.

Paul: Okay.

Perry: To collect the data, to analyze the data themselves, and to build solutions. We don’t want, you know, some jerk in a fancy office in New York City—Paul and Rich—

Paul: Uh huh.

Perry: —or in LA or in San Francisco developing a solution that then gets exported to Africa or India to say, “Hey, here’s how we’re going to fix your water problem,” or, “Here’s how we’re going to help you understand where solar energy is needed.”

Paul: So, you know, before we go to further, why is, why does this matter? And I sort of have some case studies in my head about what—you know, it seems like a lot of times we go overseas and we are like, “We’re going to fix this for you.” And then there’s, you know, just a—

Perry: Saviors everywhere, Paul. There’s saviors everywhere.

Paul: Ghost town is left behind. So articulate why this needs to exist.

Perry: Right. Well, you know, talent is equally distributed and opportunity is not, right?

Paul: Mmm hmm.

Perry: That’s a big problem. So what we want to do is, we wanted, we saw in—first of all, let me back up and say it’s our fifth birthday.

Paul: Yay!

Perry: So happy birthday to us. Five years ago when we were founded by the Rockefeller Foundation, the MasterCard Center for Inclusive Growth, what they said was, “Gosh, look at the private sector and all the cool stuff they’re able to do with data. And here’s the social sector and government, and they’re lagging behind. So how do we let them take advantage of what data lets you do? Like understanding your addressable audience and taking action that can be predictive, in addition to analytical? And how do we enable the social sector to take advantage of those tools?”

So the first thing we did was launch a big challenge which helped us understand globally the kinds of solutions that were out there. And now we’re doing a lot more work in capacity development to make sure that people in rural India are getting trained in data and AI, so that they can develop their own solutions to the problems that affect us all—those big, ugly intersectional problems like climate and health and financial inclusion.

Paul: Okay, so thematically, instead of data being something that is gathered about a group, and then somebody with a PhD far away says, “Hey, I know what you need,” we’re going to give them the tools to gather their own data, we’re going to listen to them, and we’re going to—and if they need support in analyzing and understand.

Perry: And we’re going to work with local partners who can train them in how to use the data. So, for example, we have an Indian Data Capacity Accelerator. We work with three universities—Ashoka, BITS Pilani, IIIT-Delhi—and they launch the programs that enable people to manipulate the data, to learn from the data, and to act on the data, and build software around it.

Paul: Okay, so how long have you been there?

Perry: I’ve been there about four and a half years.

Paul: Oh, so you missed the, when they said, “Gosh!” At the very beginning, you weren’t, you weren’t there. Okay, okay, okay. Most of the existence of this organization, you have been there.

Perry: Right.

Paul: It’s been a pretty eventful five years.

Perry: Yep.

Paul: Okay. Tell me a little bit about what’s changed.

Perry: The technology—you may have heard of something called AI?

Paul: Yeah, we—

Perry: It’s kind of big.

Paul: We try to avoid talking about it on this podcast, but it comes up from time to time.

Perry: Right. Well, AI is big and scary, right?

Paul: Yeah.

Perry: I think when you poll people, they say, “We don’t want no stinkin’ AI. It’s coming for our jobs. It’s coming for our families. We don’t know anything about it.” But the reality is it’s not something we need to plan for—it’s something that’s already here.

Paul: Boy, is it. All right, so you come in and it’s still the long-ago days of 2020. People are kind of, “I got this Excel spreadsheet. We gather the data, we’re going to do some stuff with it, we’re going to create a report, and then we’re going to use this locally to help doctors make better decisions,” things like that. Right?

So that was a couple of years. Now there’s this whole new thing, artificial intelligence, large language models, machine learning, and so on and so forth. And you’re trying to get that into rural or underserved places, right? You’re trying to give them the tools to deal with all this new stuff. Talk about what’s happening there. Because that is like, I don’t even know where you would start.

Perry: Well, first of all, the data still matters, right?

Paul: Mmm hmm.

Perry: The data problem, or the data opportunity, didn’t go away.

Paul: You can’t just say, “Hey Claude, make up a whole bunch of stuff for me.” Because it will, it’ll do a great job!

Perry: Well, absolutely. It’ll hallucinate the heck out of it for you.

Paul: Yeah.

Perry: But I think one of the big challenges is you still need the underlying data. You need unbiased training data, to the extent which anything can be unbiased. You need to strive for the least-biased training data. So you need to collect the data locally in a way that’s understandable, in a way that the people you’re collecting it from know what the heck is going on. So that’s super important.

Paul: I swear to God, Homeland Security is going to knock on our door and be like, “No, no, no, no, no, no.”

Perry: Wouldn’t be the first time.

Paul: Yeah, you have to let—no, it’s very un-American to say that Claude just can’t invent data about other countries. But okay, so we, so we have to have good stuff.

Perry: Right.

Paul: That doesn’t go away.

Perry: We have to have good data, and we have to have a social sector that is skilled in using that data. So we frame that as purpose, people, and practice.

Paul: Okay.

Perry: How do we think about the ways organizations are ready to use the data? And that’s everyone from people on the board who say, “Hey, this is really important, we should do it,” to how you’re working with your junior staff to incorporate data into their day to day job. So the data thing doesn’t go away.

Paul: I don’t know if you’ve noticed this, but all three of those words begin with the letter P.

Perry: Yeah.

Paul: It’s just—it’s wild.

Perry: No idea how that happened.

Paul: That is a coincidence.

Perry: It’s crazy.

Paul: Okay, so take a step back even further. The data has to be good. What does that mean? What is good data?

Perry: So good data is data that, first of all, is collected in a way that is in concert with the communities it serves. Right? So it’s not a random survey put out there. It’s collected in a way that’s culturally appropriate. So you can understand, “Gosh, why are you knocking on my door and asking these questions?” Or, “Why did this survey ask me about this?”

Paul: Any examples pop to mind?

Perry: There are a lot of questions that you ask in some developing contexts about the role of women or how much they leave the home and what’s in the home.

Paul: Mmm hmm.

Perry: And if the person who’s empowered to ask those questions is a man, you’re going to get bad data.

Paul: Yeah, no, I get this. I get this. I understand.

Perry: So you have to think about who’s the respondent versus the data you’re trying to collect. So that’s an example we’ve encountered a couple of times.

Paul: There just are different norms in different societies.

Perry: Right.

Paul: Okay. So just so, so A) I’m fully aware of actually how this culture really works and what people expect as a baseline.

Perry: Yup, right.

Paul: About gender roles or identities or religion or whatever.

Perry: Mmm hmm.

Paul: And so I’m gonna be very aware of that when I ask my questions and factor that in. So that’s one way to get good data.

Perry: Right. Another way is to keep clean data. Right?

Paul: Okay.

Perry: So make sure that you’re maintaining it, that you know where it resides, that you know it’s secure. Data privacy.

Paul: Mmm hmm.

Perry: That you’re adhering to local data-privacy concerns. So what are the data privacy and who’s the data controller?

Paul: Mmm hmm.

Perry: The big question of who’s empowered to act on the data, to use the data, in each of those instances. As we store things, you’ve heard of AI? Have you heard of the cloud?

Paul: I have, unfortunately, yeah.

Perry: So that cloud thing is everywhere. So you’ll regularly get an email from, whether it’s Airbyte or Amazon or Google, saying, “You know your stuff is in the cloud? And we said it would be in these seven jurisdictions? Now it’s in 407 jurisdictions.” So you have to think about that in terms of data privacy and data control and when you’re reporting back, like, how your data is being stored and where, that’s an important consideration.

So a lot of complexity around data. That’s pretty much good data. And good data is data you understand, right? And I always say it’s sort of the right amount of data. I mean, I’m a marketer by training, and marketers are like, you know, “Where’d you get that hoodie? What size is it? Where are the cables from?” We ask a lot of questions. Sometimes we don’t really—

Paul: “Why’d you click on that ad?”

Perry: Exactly.

Paul: “Would you like Andrew Cuomo to be the mayor?”

Perry: Yeah.

Paul: Those are the things—

Perry: “Would you recommend this operating system to a friend?”

Paul: [laughing] Yeah, I love, I love that question. When that was, it was Twitter back in the day, just started dropping those all the time. It was like, “Hey, what do you think about Windows XP 11 1.5?” Or whatever. [laughter] It was just so bananas. Even though we live in a magical world where you can say, “Hey Claude, do this, that, and the other.”

Perry: Right.

Paul: So you have to go out and you have to really be aware. And honestly this is true if you’re talking to middle-class people with Betterment accounts, too.

Perry: Right.

Paul: It’s just that we can take more for granted when it’s when you’re right there. So, okay, so people who know the community must be those who gather the data. And they need to be sensitive to what, how community preferences and biases, not external biases, but the local ones, are going to change how people react.

Perry: They have to get the why and the what.

Paul: Okay.

Perry: Why do you want this information, and what are you going to do with it?

Paul: Okay. Because I don’t think most people on earth have ever collected data. Right?

Perry: Right.

Paul: Like, I think you’re a marketer, that, like, that’s your background.

Perry: Yeah, we just want to ask a few questions.

Paul: Yeah, exactly. And so, but also that is how you justify your existence.

Perry: Right.

Paul: Like, I’m going to go spend all this money on marketing, or on, we’re going to build a new building or whatever, and we need to know what people are going to do if we actually do it. We need to understand where those millions of dollars are going to go. So I think, like, your world is very data-driven. But I don’t think, you know, most people don’t go out and collect data as a way to get the day going.

Perry: Most people spend their days being a data point, right?

Paul: That’s right.

Perry: That’s the scary thing is we feel like we’re on the tail end of the data all the time. You go to the vet, they want to know, “How was your experience?” Every single interaction we have right now is surveyed and acted as data.

Paul: Do you think that, just a little digression, does that matter? Does anyone really care? At a certain point, it’s so much. I get polled about everything now.

Perry: Right.

Paul: And sometimes I do it, sometimes I don’t. Usually I don’t. Like, does it just get pointless at this scale and there’s just a bunch of blobby data everywhere.

Perry: Yeah. I’ll go back to your point of like what’s good data? And I always answer the right amount of data.

Paul: Right.

Perry: Right? Don’t ask questions where you’re not going to act on the answers. Don’t ask questions that are just going to wear out your audience, in hope that someday, this Net Promoter Score will make a critical difference in your career or life.

Paul: You’ve been doing this for a long, long time. Well, no, not, you know, weeks. [laughter] But the—sorry, that’s a little—

Perry: Since dinosaurs roamed the earth.

Paul: Little Gen-X sensitivity right there. That’s all, that’s all I, that’s all I was doing. But you have been at this and you’ve been at it at high levels. You’ve worked at places like Harvard University, which I don’t know if anybody’s heard of it, but it’s a, it’s a, one of the—

Perry: Another small not-for-profit.

Paul: All right, so here you are, you’re somebody who might put out a poll or… You are in a data-driven career and you’ve made, I know this about you, you’ve made very big decisions involving lots of money around data. Like, okay, this is what it shows us. We’re going to do this.

Perry: Right.

Paul: We’re going to market this way. We’re going to build these things. Hire these people. What is the role of anecdote? How do you weight a good powerful anecdote in all of that thinking?

Perry: You got to have head and heart, right?

Paul: Yeah.

Perry: You need to have the data that tells you that the direction of travel, you know, really helps you validate some of your assumptions, helps you counter some of your assumptions. And actually, when we talk about the role of data in the social sector, I think one of the big challenges of getting people to adopt it was people had strongly held beliefs.

Paul: Mmm hmm.

Perry: If you give kids school lunch, they’ll do better in the afternoon.

Paul: Mmm hmm.

Perry: Whatever that belief is. And when we start to use data to interrogate these really passionately held points of view, you’re going to find some of your assumptions were misses.

Paul: Mmm hmm.

Perry: And that’s a lot easier to do if you’re selling widgets, you know, from a factory than if you have a mission-driven not-for-profit, where data is super important.

Paul: All right, so we have to, we have to stop feeding the children.

Perry: Absolutely. No food for any children.

Paul: No, that’s—thank God. It’s good, it’s good—and that’s—

Perry: Because then they just expect it, right?

Paul: Yeah, no, and then they’re constantly like, “Let me just have more food.”

Perry: Yeah.

Paul: And they put their little porridge bowl out. [laughter] It’s just horrible. Don’t do that. So, okay—

Perry: So that’s head and heart. The head part is we need the data. We need to validate our assumptions and we need to find out some new stuff. And data will do that. But if you want to be memorable, you have the anecdote. That’s the heart, right? That’s the story you tell on top of the data. So if I’m ever delivering to a decision-making team, you know, what, what the data is, you need to have a couple of stories that illustrate what that data represents or means.

Paul: I mean, if there’s one piece of advice that people could pluck out of this conversation before we even get to the AI part?

Perry: Right.

Paul: It’s actually that.

Perry: Well, let me go back to the Inclusive Growth and Recovery Challenge.

Paul: Mmm hmm.

Perry: So one of the examples we got there was an organization called BASE, which is cold-chain storage for smallholder farmers. That’s a heck of a lot of jargon in one sentence, but.

Paul: So cold-chain is like, I need my fish to be cold.

Perry: Exactly. Like, keep my stuff cold.

Paul: I’m not being metaphoric, by the way, like—

Perry: Yeah, exactly. Like, literal stuff.

Paul: Okay.

Perry: The data showed that there was a huge need. That, you know, farmers were coming to the market with their fresh vegetables, and on day two, day three, you can’t sell them as well because they’re not as fresh.

Paul: Mmm hmm.

Perry: And there was a lot of potential, you know, livelihoods lost.

Paul: Sure.

Perry: Because people were unable to maintain it, especially—you’ve heard about climate crisis, you’ve heard of that?

Paul: It comes up from time to time.

Perry: Okay, so AI, cloud, climate crisis. I’m catching you up on all this stuff.

Paul: Thank you.

Perry: But on the climate-crisis piece, you know, there, there are parts of the world are getting a little warmer.

Paul: Right.

Perry: So this is a real crisis if you make your livelihood, you know, selling fresh vegetables. So cold-chain storage—

Paul: Tomato trouble.

Perry: [laughing] Exactly.

Paul: That’s two Ts.

Perry: Yeah.

Paul: Yeah, there we go. Tomato trouble.

Perry: So climate crisis, smallholder farmers, and they get cold-chain storage to make it better. So it’s one thing to say, you know, that jargon, very jargony sentence I led off with, and it’s another thing to picture the farmer. Right? To think about that one person in, started out in India, they actually were able to port the solution to Nigeria, which is a big Data.org principle. We want to be locally led but globally informed, right? We want to do stuff locally that’s really smart, and then be able to port it where it makes sense in other cultural contexts. So this is a project that started in India and then moved to Nigeria. So when I was in Nigeria with two of my colleagues, I said, “Hey, let’s go across Lagos,” which is, you know, just a hop, skip, and two and a half hours in a car.

Paul: Yeah. Little city. Quaint.

Perry: Exactly. To go check out this, you know, organization that, with MasterCard and Rockefeller, we were able to fund.

Paul: Okay.

Perry: And we went over there and checked out and we felt the green beans and we talked to the farmers and we talked to the guy who ran the storage and we understood the role of data—

Paul: Were they cold? Were the green beans cold?

Perry: They were coolish.

Paul: Okay, that’s good—

Perry: You know, I wouldn’t say they were arctic, but they were coolish.

Paul: I actually don’t, you don’t want your green beans too cold.

Perry: Exactly.

Paul: Let’s be honest here. You want them, like, that nice, firm green-bean texture. So that’s a global thing. Okay.

Perry: Yeah.

Paul: So we can, I mean I’m going to imagine that the green-bean density is actually pretty global, that everybody wants a nice cool but not frozen green bean.

Perry: Yeah, but just being able to, like, see the green beans and talk to the guy who runs the storage and see the farmers come up and transact and store their, you know, wares there overnight and be able to come back the next day. And data made that happen, right?

Paul: Okay.

Perry: Understanding where to put those cold-chain areas and then using data to understand how will people use a smartphone in order to be able to interact. Like, one of the early assumptions of that project was the individual farmers would have a smartphone. Mmm, not so much.

Paul: Mmm hmm.

Perry: But the person managing the cold chain-storage did.

Paul: This is all very cool, but it strikes me as extremely late 2023, where as I’m—

Perry: The world has changed. We’ve moved on.

Paul: I’m very 2025 these days. So I’m, like, “Eh, green beans, data, whatever.”

Perry: Right.

Paul: “Where’s Claude? Where’s ChatGPT? Where’s Sam Altman in all this?”

Perry: Yeah, well we’re all about AI now, so. [laughter]

Paul: So okay. I mean, if you, you kind of can’t be an organization with the URL Data.org and not be deeply sort of connected to some of the changes. So yeah, talk through a little bit of that. And I’m really interested, actually, because you know what strikes me with your org is that there’s a lot you reject and some you take in. Like, you’re careful about which parts of our big world of data science and visualization, because there’s just so much froth around data—

Perry: And there’s so many people doing great work.

Paul: Yeah.

Perry: Right? You want to carve out your piece of it.

Paul: That’s right.

Perry: Ours is really about capacity and challenges. Like, how do we get people smarter about this stuff, and how do we find these great use cases, these great stories—to your point of, where is anecdote? Well, the challenges are engines of discovery. They help us find that anecdote, help us find these exemplars of people doing great stuff with data, and today, since it’s 2025, they’re doing it with AI.

Paul: Okay, so we had green beans, okay? And now we have Claude. What’s, what’s changing? How is that going to get out into the world? This, this new thing?

Perry: First of all, for the sophisticated green-bean data technology, you need to have a means to collect the data. That’s where we started this conversation. Store the data, clean the data, analyze the data.

Paul: Mmm hmm.

Perry: And now we have AI, which is kind of a ground-up movement. As much as we’re all saying, like, “It’s terrifying and I don’t want it and why does everything say AI inside, from my Coke can to my, you know, my email software, is telling me all about its AI. I don’t want to hear that.”

Paul: Yeah.

Perry: But I am seeing unbelievable examples of people using AI from the ground-up within organizations, and it may be a superpower for the social sector.

Paul: Yeah. All right, so give me some examples. Tell me what’s going on.

Perry: So within the social sector, I see huge opportunity for people just to understand what the heck is going on better. I mean, the tools around summarization of reports. You know how many reports come out around data and AI in the social sector? There have been 19 since we started talking today.

Paul: God, no, I do actually. That’s the thing. Like, every now and then I’m like, “Oh, how’s—oh no.”

Perry: “I’ll dive in.” Exactly.

Paul: Yeah.

Perry: “Here’s a 45-page PDF for you, Paul.”

Paul: Just burn my RSS reader. It’s great, though, that we spend so much money generating those reports. It’s exactly the right prioritization.

Perry: Well, the thunk when it goes on philanthropy’s desk is, you know, definitely definitive.

Paul: Yeah.

Perry: It does enable people working the social sector to have greater mastery of the, the new things that are coming out there. To try new stuff, to build stuff. And that stuff is emerging. And one thing I love is seeing how people are able—we recently ran two challenges. One was Microsoft’s Generative AI Skills Challenge, really focused on upskilling and reskilling toward AI. And the second was with MasterCard, focused on AI—to AI Challenge to Accelerate Inclusion. Basically, how are we going to use AI to get people more financially included, healthy members, productive members of society?

Paul: So tell me how, tell me like I’m a small child who—or, like, or a squirrel.

Perry: Right?

Paul: Okay.

Perry: So let’s start with the Generative AI Skills Challenge.

Paul: Okay.

Perry: So we found an organization called the Myna Mahila Foundation, based in India. And what it does is leverages large LLMs—does not build them, right? Takes advantage of LLMs that are out there.

Paul: Yeah. Don’t build your own LLM.

Perry: Exactly.

Paul: Anyone. Everyone should know not to do that. Okay.

Perry: So in order to provide rural women with access to women’s health information.

Paul: Okay.

Perry: To have conversations they would not have, you know, with maybe their sister or their cousin or their doctor or their spouse.

Paul: Oh, interesting. So it provides a kind of conversational privacy.

Perry: Exactly.

Paul: Interesting. And they know that it will be safe and secure.

Perry: Yep.

Paul: Okay.

Perry: So you’re able to, you know, speak into it and get answers out of it. And there’s a human in the loop, right? Vetting the answers, checking for hallucinations. So, you know, you never want to turn it over completely. But there are solutions that are being developed, and I think those are, those are the heart stories. Right? Those are really tangible examples of, “Oh, I see that that’s a real problem.” This is not a technology you want to go out and, you know, chase down a problem statement. They’re real problems that you could just apply this technology to today.

Paul: Okay, so I see technology, in our culture, in American culture, and it’s like they’re just about just, the current administration is talking about, like, we’re going to put $500 billion towards AI data and so on and so forth.

Perry: But spend it efficiently.

Paul: God help us. And so, like, basically the deal is, “Hey, I want innovation and I want things to move forward. And the way that I’m going to do that is I’m going to actually point a fire hose of money at people who already have a fire hose of money. And I think innovation will come out.” And that is sort of the American deal. And it often works, right? And so that’s our understanding and sometimes my understanding of how tech works. Meanwhile—

Perry: That’s the AGI Moonshot.

Paul: Yeah, that’s right. That’s right. And sort of whatever falls out of it, we get to take credit for as well.

Perry: Right.

Paul: Meanwhile, China is making its own LLM, DeepSeek, which it generated—it was able to do kind of the work of OpenAI, but for maybe a hundredth of the cost. It’s 30, 40 times cheaper.

Perry: Mmm hmm.

Paul: They used the crappy chips that we let them have. And so you can see, like, okay, that’s a totally different approach. Right? Like, we, okay, I don’t have the resources, then fine, I’m going to innovate based on, based on the constraints. But still, like, this is a very, China has as many GPUs as it needed. It did a lot of, you know, some of them were, had been used for mining—

Perry: They could throw a few folks at the problem.

Paul: Yeah, exactly, right? So now, so that’s—so we have the tons of money, as few constraints as possible. Lots of constraints, lots of resources. And now we’re in your world, which is really very few resources and tons of constraints. I have a computer. I have access to the internet. I’m assuming those are kind of my two basics.

Perry: And also constraints around responsibility, right?

Paul: Okay. What does that—wait, what does that mean?

Perry: So thinking about, are we doing this, you know, responsibly? Are we thinking about the potential repercussions on peoples’ lives?

Paul: You don’t think Sam Altman is sitting there worrying about people in India?

Perry: There may be a few folks over there worrying about it, but we worry about it as a first and foremost, right?

Paul: Okay.

Perry: Are we doing the right thing?

Paul: Okay. And so what is innovation? What is tech like in those environments, compared to here? Like, when you are in Lagos and you’re watching, is it just kind of the same, but they have less money? Or like, how—how do people see this industry and this space?

Perry: I would say there’s tremendous power in constraint.

Paul: Yeah, okay.

Perry: Right? You can find people who are innovating in remarkable ways, whether it’s around AI or sneakernet, right?

Paul: Mmm hmm.

Perry: They find ways to solve important and pressing problems without excessive debate. So I would say some of those solutions we’re seeing—well, I can take an example, you don’t have to go to Lagos. You can stay right here in our own country. We, through our latest challenge, funded an organization called Link Health. And what it does is matches people with federal assistance programs that they qualify for, no net-new dollars, right? The money’s already allocated out there.

Paul: Okay.

Perry: But it uses AI to predict where people might have these needs, and connects people with the resources that they need.

Paul: So this stuff is landing as a layer in your world. The data still has to be clean.

Perry: Yeah.

Paul: So Data.org isn’t changing its mission, which is sort of get, get—

Perry: AI.org was taken.

Paul: [laughing] That is a… It’s a good URL. I can’t imagine—I don’t even want to go look at what’s there.

Perry: Right.

Paul: You still have to go talk to people. Got to get them to tell you what they need.

Perry: Data still matters. Digital still matters, Paul.

Paul: Okay.

Perry: I mean, one of the things we’re hearing from funders more and more is, “We want people upskilled on AI. We cannot afford to wait.”

Paul: Okay.

Perry: “We can’t sit this one out. No one can sit this one out.” Data still matters, but there’s a huge swath of the world still trying to get access to Excel, right?

Paul: Yeah.

Perry: And we have to worry and think about all those folks. Now, to what degree? A really interesting question is, will the AI obviate the need for the Excel, right? To what extent will empowering people with AI let them leapfrog, like, the cellphone networks, right? To get over the landlines?

Paul: Right.

Perry: Is there something there that’s going to happen with AI?

Paul: This is how I feel in general, too. I mean, I’m just like, I’m not really a global thinker and I wish I was more of one. But as someone who is very self-taught and really wanted to understand technology kind of from first principles when I was a kid, and it took us a while to afford a computer and all that stuff, when I watch new technology show up, in a way, like, blockchain, I kind of rolled my eyes at. And maybe I was, I mean now it’s, you know, $2 trillion in the economy, so maybe I shouldn’t have.

Perry: You’re not an NFT billionaire?

Paul: I’m really not. But with this one, my instinct was not, “Oh my God, I want that.” My instinct was, this is going to change a lot of stuff, and we need to make sure everybody has access to, to it and understands it. And what I saw was a lot of people panicking about generative and getting really angry at it, which I—often for really, really good reason. But I’m, like, you can’t put it back in the box.

Perry: Well, it’s like the early Wikipedia reaction, right?

Paul: Yes.

Perry: Do not let your kids cite Wikipedia, right?

Paul: No, exactly.

Perry: And you know, we’re all on a journey. But I agree, people have to get their hands dirty. They have to experiment with it. So the wildfires in California.

Paul: Yeah.

Perry: I saw so many interesting things pop up of people developing solutions for—step one was, like, “ChatGPT, how do I prepare for a wildfire? There’s more fires in my region. How do I prevent, how do I protect my house?”

Paul: Mmm hmm.

Perry: Step two, I’m seeing relatively non-technical folks work with tools like Claude from Anthropic to say, “How do I assess my insurance risk of my house?”

Paul: Sure.

Perry: You know, “How do I prioritize the things that I do in order to protect my home or to work with my insurer?” And you’re seeing applications, full-fledged applications pop up in the weeks since the fires.

Paul: Interesting. Okay, so it’s an accelerant.

Perry: Right.

Paul: Fast forward. Let’s say, don’t assume any magical, new wonderful AI super robot, but just kind of here we are now. It makes mistakes, think—but you can accelerate certain things like writing code, doing certain kinds of research. It’s good at conversational interfaces. Where’s this headed in your world? Okay? I mean, you’re doing things like helping people distribute solar ovens or, you know, increasing health outcomes. And so, and so like, is it just kind, of things are just going to move a little more quickly? Do you have a, do you have a sense in your head of where you, where it’s going to go? Or are you just sort of like the rest of us going, “I have no clue.”

Perry: What’s the quote? Gradually and then suddenly?

Paul: Yeah, yeah.

Perry: So I think, I think that’s where we’re headed. I think we’re now in the—

Paul: No, no, this can’t be gradually. This has to be suddenly.

Perry: Yeah.

Paul: I’m not ready for suddenly to turn into more suddenly.

Perry: Yeah, no, I think, I think people are hoping it’s gradually, but it’s actually suddenly. [Paul sighs] So I think people are really looking at what are the tools we can use, need to use, that’ll change—and I think it’ll happen from the ground-up. So when we started this conversation, we were talking about way back, kids gather around the fire, because way back when you needed to start a website, you had to buy a Sun server and you had to hire a webmaster and there are all these steps you had to do.

Paul: Hell yeah!

Perry: And now you got Squarespace.

Paul: Yeah. [disappointed noise]

Perry: [laughing] But looking to today, I think the same thing is going to be true with AI, right? We’re going to see people within organizations adopting and adapting quickly to using AI tools to do their jobs.

Paul: It’ll just kind of be the substrate. So really, every project that Data.org gets kicked off is going to have a layer on top, or some, some connection to—

Perry: Or under the hood.

Paul: Right.

Perry: You know, I think one of the mistakes we’re making about AI is sort of like Fight Club. We’re talking about AI.

Paul: Mmm hmm.

Perry: I think we need to be transparent about where AI is used. But I think so many private companies and projects are just, you know, branding everything “AI.” Whereas what we need to do is focus on what’s the problem statement? What’s the thing we’re trying to fix?

Paul: No! That’s the least-American thing I’ve ever heard in my life.

Perry: [laughing] And then figure out how AI is part of that solution, if needed. Like, where does accelerate? But I do see the individual usage being a big driver here rather than the institutional adoption, particularly in the social sector. I think you’re going to find smart people who gravitate toward these tools, who use them in other contexts to run their social organizations, their kids’ sports teams. And you may have heard this, Aboard, but it’s quick and easy to spin up software these days. It’s not what it used to be. It’s kind of requirement-less.

Paul: What is—or “reqless,” as we say. What—what are you doing with it on a day-to-day basis? Does Perry have applications for these new tools?

Perry: Summarization is a big one.

Paul: Okay, okay.

Perry: You know, so to help me understand what is going on.

Paul: You haven’t heard a word I’ve said. You’re just waiting. [laughter] Okay, so summarization.

Perry: That’s a big piece of it. Also, you know, brainstorming. You know, I’m surprised to degree to which in my personal and professional life I use it to bounce ideas off, you know, what are some of the ways that, you know, if I need to solve a particular problem, if I need to meet up with somebody in a foreign city, what are the things I need to do, think about, get connected around?

Paul: It’s extraordinarily useful for context shifts like that.

Perry: Right.

Paul: I think about it from back in, I’m not selling services like I used to in my old life, or working as a journalist, but it always was really hard and really alienating to figure out where the other person came from. Like, just sort of what they were about. And so it’d be like, you know, I’d have to go interview an art dealer or something like that. And if I could have said, “Hey, tell me the things that art dealers tend to care about.” I don’t want it to give me the questions for the interview I’m going to perform as a journalist. I just want it to tell me kind of what they’re, what it thinks their worldview is. Because I’m going to be 30% more likely to get a good outcome. And so yeah, I think about it a lot.

Perry: Well, you and I have both had careers as lucky generalists.

Paul: Yeah, that’s right.

Perry: You’re far more technical. But you know, we write a little, we think a little, we research a little. And I do think when you’re a lucky generalist who works on a lot of different things—I’ve worked in the education sector, the for-profit software sector, and I work in a different global context. And when you go to other countries or you need to meet with different kinds of people, AI is like the ultimate cheat sheet.

Paul: That’s right. Or as I like to call myself, a very shallow thinker. But yes—but broad!

Perry: Mile wide, inch deep.

Paul: Exactly. Okay, so I know I can go to Data.org because it’s literally a URL. Okay, so that part, you’ve done that. From a marketing point of view, that part’s really good.

Perry: I got a brand that was born on third base, buddy.

Paul: That’s right.

Perry: Yeah.

Paul: That’s very, very strong. So, but what else do you need? What else should people look at? Tell the people what to do.

Perry: I think people in the social sector, in terms of data and AI? In terms of AI, get your hands dirty, get involved.

Paul: Yeah.

Perry: Give AI a go, figure out what are the tasks you can use it for to advance your day-to-day life, and then also think about what is it going to do for your organization. Because if you’re not worried about that, someone in your organization is already doing it. So you got to catch up and get involved.

For Data.org, i think we’d love you to come to the URL, come check out the site, see the case studies, see the exemplars. And when you look at things like our challenge awardees or some of our capacity programs, you see, like, we want to paint the art of what’s possible. We spend a lot of time envisioning our AI dystopian reality. And I, too, sweat when I see the robot dog sprint. Like, it’s scary.

Paul: Yeah.

Perry: Like, I get it. That said, there’s really a lot of good that’s possible. And if we spent 1/100th of the time that we do perseverating about all the terrible things AI will do to society—which P.S., we should worry about and regulate and think about as well—and we looked at some of the ways that constraint and creativity and passion are driving people in the social sector to use data and AI for good? I think painting picture of what’s possible for others is going to put a lot of hope and hopefully a lot of positive momentum out into the world.

Paul: That, I mean, you know, I went into BlueSky today and the first thing I saw was, “If you use AI, you’re supporting fascism.” Which I was, like, “Maybe?” But that’s real, right? There’s such a counter-movement to this, and my reaction to that isn’t like, “No, no, I’m okay, you don’t have to worry about me.” My reaction to that is, “Sure, that could be true, but frankly, if I don’t understand it, I can’t actually decide if that’s real or not. I need to understand.” And so I think everybody kind of does owe it to themselves to just try to figure this thing out.

Perry: I mean, whether it’s technology, data, or AI, we don’t want it done to us, we want it done with us and alongside us. We cannot put the genie back in the bottle. AI is here. So what are some of the great futures we envision? What are some of the problem statements we know we need to work on? Climate, health, financial inclusion. Let’s work on those problems and figure out how technology can be used to serve us, not to be done to us.

Paul: Lot of genies coming out of bottles these days. So on that note, Data.org. You can look up Perry on LinkedIn. She has a very nicely curated profile. It’s lovely.

And you’ve been listening to Reqless—R-E-Q-L-E-S-S, the podcast of Aboard. You know, I forgot to, I’m talking to a marketing leader and I forgot to market my own company. Your regular co-host, Richard, and I are the co-founders and Aboard is a tool for building software really, really quickly. It is AI-powered. If you go to the website and you go look at the top right, there’s a button that says “try it.” Click that and try it, and you’ll see that you can actually—we’re getting to this point where you can describe what you need and it’ll just kind of build it for you, and it’s weird and really exciting.

So go check it out. If you need anything at all. Hello@aboard.com. And we will talk to everyone soon. Thank you so much for listening.

[outro music]