Welcome to CDT's Tech Talk, where we dish on tech and Internet policy while also explaining what these policies mean to our daily lives. I'm Jamal Magby, and it's time to talk tech.

CT. Team. Today, we're diving into a critical conversation about bias and equity in AI systems. Joining us today are two incredible guests, Rafi Krikorian, chief technology officer at the Emerson Collective, and Alexandra Givens, president and CEO of the Center for Democracy and Technology. Let's get started. Raffy and Alex, it's so great to have you both here today. Thanks for having us. Thanks for having us. Of course. So, Raffy, can you kick us off and talk a little bit about your background? You joined Twitter back in 2009, then you went on to work at Uber, then the DNC, and now at the Emerson Collective. What what's the thread that connects these roles, and what drives your work in tech? Yeah. It's a great question. I mean,

I have been lucky. I've been in a bunch of fun places. For me, I think it's two things. One is I'm always in a pursuit of learning something new. So, like, I joined Twitter because it was a social network that had dreams of flattening the world. I joined Uber, specifically self driving cars at Uber, because we had an opportunity to reenvision how cities work. Went to the DNC because I felt the world was broken, and maybe there was something I could do to help, and then continuing that work at Emerson Collective. So for me, it's always been I'm trying to learn something new. But as you can hear, there's like a through line of I I just wanna help people. Like, I feel fundamentally, I believe that technology can be used to help people in a bunch of different ways. You know, my grandfather used to run a radio station long time ago in a place called Cagayan De Oro in The Philippines. And, you know, he would have farmers come to the radio station at night just to broadcast they were gonna go home. So, like, their wives and families knew they were about to start the trail back to home. And for me, that's always been like, you can use tech to do great things. So I'm I'm just in a search for ways I can use tech to help in those kind of ways. God. I love that vision of connecting people.

Can you talk about that early time and kind of your reflections on the company? I mean, I you're right. I mean, I feel like in a lot of ways, I grew up in that company at that time period. Like, you know, we were just I'd started just around the time the Arab Spring had finished, so we're sort of feeling all that intense ability that, like, Twitter can be used to shine both problems that we see in governments. But also, you know, around that time, there was the Fukushima earthquake. And I remember being awake at night, Random reason. No no particular thing to do at work. But, you know, being a workaholic, I just had my laptop open, and I started seeing the our graphs of how the Twitter health Twitter health service was going, how many tweets per second were going, just starts spiking for no good reason. I couldn't tell what's going on. And I I immediately Slacked, I can't remember what we used at the time. Let's call it Slack. I immediately Slacked, our ops team being like, what's going on? They're like, turn on the news. So, like, literally, you can see world events Yeah. Occurring in the Twitter the Twitter graphs in the Twitter system. And all the way down to, like, kids would be celebrating their first prom on Twitter. Right? So, like, you could have this whole zoom in, zoom out level. So for me, you know, again, super idealistic at the time. Like, we clearly didn't have a great handle on all the crazy ways something like this would go. But there was just like a celebration of life and, like, a celebration of of community

Yeah. I mean, so at Twitter, I was fortunate to be you know, I can't remember the exact number. Maybe let's call it employee 50 at the company, but I was the first person the first engineer hired on what was called the platform team. And our job was to build the Twitter API. Like, how do we have the world integrate themselves into Twitter? So this is the days before there was a Twitter mobile app. We only had a Twitter website. And so me and a friend, Ryan Sarver, I would be the engineering lead. He would be the product manager. We worked with hundreds of developers who are all building their own mobile Twitter apps, their own mobile integrations, their own whatever it may be to plug into the Twitter stream and contribute to the community as well as find insights and analytics from what was going on. So that Twitter for me was sort of like this, like, real engineering challenge. And I rose to become the VP of engineering, specifically in charge of global infrastructure. So, like, how do we build Twitter to be scalable, reliant, efficient across the globe? So, you know, I remember the two thousand and I wanna say two thousand and eight World Cup, which was in South Africa, and we would literally be hoping no one would score a goal. Like, every time someone scored a goal, Twitter would go down. Like, so many people retweeting, it would just go down. And so for four years, I built up, you know, a 500 person team worldwide so that when when the World Cup happened again, I think it was in Brazil the next time, there was you couldn't tell. Like, there was not a blip in the service because we had we had solved all the big infrastructure problems. I then moved over to Pittsburgh, Pennsylvania where Travis Klacnick found me and asked me to help him start the self driving car team out there. And I knew nothing about robotics, but I knew lots about deploying large scale systems. So for me, it was learning how to, like, work in this, let's call it research to development continuum. Like, no one actually knows. Maybe today, we know how to build a self driving car. But back then in 2014, no one had any idea how to build a self driving car except maybe Waymo, but they still hadn't figured it out yet. So I stole a bunch of researchers out of Carnegie Mellon, and we, like, put engineers around them, and we desperately tried to get algorithms working, training systems working. We built a test track out in Butler, Pennsylvania. And within a year, we had a car that plausibly drove on the road. It picked me up every morning. And when my my kids would go to the wind and be like, baba, robot car is here. And then, like, I would get I would get in the car, cross my fingers, and then they would drive me to work. But trying to figure out for me, that was a management leadership r to d challenge. We're trying to figure out how do we take cutting edge research and actually put it in a life critical situation. Then, like you said, I moved to the DNC, and there, you know, I was used to being at the the big technologist at a big tech company. Like, everyone wants to talk to you. It's like everyone is knocking at your door. What's your road map? What are you doing? What's the big feature with I love where this answer is going, by the way. Yeah. And then at the DNC, I'm, like, literally knocking on every door being like, can I join this meeting? Like, please? I have some value add here. Yeah. I mean, I I I I promise I won't mess anything up. I did mess plenty of things up. But I was just sort of, like, trying trying to figure out how to get into political meetings, trying to figure out how to get into meetings with elected officials just to be like a different person in the room. The technology team at the DNC is now actually the largest team at the DNC. There's over 60 technologists, at the DNC building things around the voter file, building things about ways to get voters to turn out and vote, how to track their early vote status, doing cutting edge analytic. For me, that was, like, really reforming the structure of that organization because the Democratic Party is the world's longest running continuous political party. And so as you can imagine, it kinda looks like the eighteen hundreds in a bunch of different ways. So trying to get them into the twenty first century as quickly as possible, which not just meant technology, it also meant technology investments. Like, an iPhone shows up after decades of innovation, not in one year of working. So, like, how do we set them on a trail that we can get to that place? And now I get to work at the Emerson Collective where I get to think about broadly this intersection of tech, society, humanity, and how do we just, again, try to find opportunities to use technology for the social good? How do we educate policymakers so that they can make a more intelligent choices when it comes to how technology is used and viewed both positively and negatively? And then how do we just make a difference in the world? That's what the thing that my team is always searching. What's the biggest difference we can make? I'm I'm happy we ended there because in 2023,

you jumped into the public debate about AI. You wrote, and I quote,

we need a more balanced approach to the public debate around AI and regulations that governs it. How do you view the conversation about AI governance right now? What should policymakers be focused on? Yeah. So I think about two different things. Maybe I'll plop two ideas on the table, and we could talk about that. One is I I'm I think a lot about these, like, I call it the three legged stool. Like, right now, like, there are commercial actors, there are governmental actors, and there are civil society actors. But in order for us to actually really get to a great place, we need to get the three legs of that stool to be in a balanced place. And right now, we are clearly imbalanced. Right? Like, a lot of this is driven by one of those legs, the commercial actors. There is, like, burgeoning desire on the governmental side, but the social sector almost is absent from all these conversations. Not completely absent, but almost absent. And so I think a lot about how do we, like, try to get this other two legs of the stool up and running. So, like, my my podcast, my newsletter has all all been around trying to get those other two legs of the stool activated in some way. I don't know if it's working. In fact, in our lots of ways, it's really hard. So but that that's the thing I'm trying to get done. And then the other thing I would note is that especially in the philanthropy sector where I spend most of my time these days, maybe a small portion of that civil society, we talk a lot about harms mitigation or if I say more crassly shit mitigation, like, especially when it comes to AI. And I don't think we do enough talking about, like, optimistic futures. I don't think we do enough about thinking about opportunities because I feel a lot of people when they think about opportunities, they feel conflicted of just like but I also worry about the harms. So I spent a lot of time in this sector trying to be like, we can think about both. Like, both of these can be true. It could be awesome, and we have to mitigate a bunch. So those are the two big ones at least I'm trying to spend a lot of time focusing on. And then I love your second point about kind of this affirmative vision

one, just so for full disclosure, I'm actually on the board of the Mozilla Foundation. So I think a lot about how this moment in time looks a lot like when web browsers first showed up and how access to information just showed up. I think that the public act the public interest public interest artificial intelligence, I think, has, like, two different aspects to it. There is one which is we need to tell clear stories of what does trustworthy artificial intelligence look like? What does it mean to have visibility into bias, into decision making, into all these things around how these big systems are being developed for us today? Instead of us necessarily having systems that are developed purely in the commercial interest, what would it mean to actually build systems that are with and by the community in a lot of ways? So it's our data that's being used to train these systems. How do we understand what that looks like? How do we get more instead of treating them as black boxes, how do we treat them as clear boxes, for example? But I was actually speaking as sort of, like, even a bigger level of, like, I actually feel that there is this I'm borrowing a a phrase from a friend of mine, but I feel like they were in this middle of a crisis of a lack of imagination. So, like, I feel like this public interest artificial intelligence and public interest technology, it's such a wonderful thing. We need to do it. More technologists need to be trained not in questions of ad building systems when they're in college, but also in questions about how do we use technology to work on homelessness or food insecurity, things like that. But for me, that actually still fits into my bucket of harms mitigation. We also need to be talking about a world of, like, what would it mean to have abundance in a bunch of ways around artificial intelligence. And so I feel like there are these narrative moments that we can latch onto. You know, for example, for all the nerds out there, you know, in the nineteen sixties, nineteen seventies, Star Trek talked about tricorders all the time. Now DARPA literally has grants asking for people to create tricorders. Like, how do we get back to that kind of imagination, especially when it comes to technology and artificial intelligence? What's the right narrative path so we can really inspire a generation of people to, yes, we need to work on climate change. Absolutely. Yes. We need to work on on insecurity. Yes. We also need to have people have ownership over their data and their destiny. But we also need to think bigger about all these technologies.

Such a great question. The market does not get us there, in my opinion. The market gets us to you know, the market currently if we go back to my three legged stool analogy, the market is very tilted toward commercial incentives and short term monetization. Absolutely makes sense. That's how capitalism works. Government might get us there. You know, things like ChipX was great as a way to try to figure out how to get us to actually invest in these type of things. I would love the government to shift their mind frame from shit mitigation to also inspiration. But I think it also means that we need to get regular people, kids excited about what these possibilities look like. And look, for artificial intelligence specifically, we spent my entire childhood, if not longer, talking about Terminator and talking about all these other that's the narrative. And so we need to just figure out what is that positive narrative so my kids, our kids can all be thinking about these opportunities looking forward. And we could just all dream. Because if we can marry that kind of dreaming with notions around public interest technology, then I think the real magic happens in our society.

like, she did this she did this, this competition, this call to action where she tried to get people to describe what the Internet should look like for different types of people. Like, what would it look like for a grandmother in Liberia? What would it look like for a a rural a rural child in India? Like, what should the Internet look like? Because right now, like, the imagination only comes from the portion of the planet that you and I are sitting in right now. Like, how do we just activate everyone? Because I feel like there's so many great ideas out there. We just need to bring it all together, and then it can all come back down to building systems that actually matter for all of us just with a broader lens of what's possible.

Yeah. I mean, I think we need to I'm so glad you said that. I'm so glad that's the orientation that you are all coming from because I think that we need more people to realize that these are nuanced things. Like, it's like we all look for the black and white, and we just got to be comfortable in the gray. Like, we're going to be making a hard trade off. Like, the trade off I think a lot about these days is the open source trade off. Like, I mean, I'm a strong open source proponent. I really feel that that is the way we get to trustworthy systems. But, you know, my colleagues at Stanford Law will hammer me every single time I mention it of, like, but these systems are what's gonna create an endless stream of CSAM on the Internet. Now it's just like, I I understand. I mean, look, I'm not an asshole. I I I obviously think that's an incredibly bad thing, but these are trade offs. Like, we have to think about what's in the greater good in a lot of ways. So I I would love to find the solution to do both, but in it's not in the world that we can't, we have to make some, like,

Your episodes explore the exact issues that drive CDT's work from season one, which focused on the effects of AI on our society and the approaches to regulating it, to your season two deep drive and data, from targeted and behavioral advertising to election integrity, to social media platforms and kids' safety, to reproductive privacy, to facial recognition, to data driven policing. You literally had an episode

it was really great to see you putting that attention on that issue. Oh, thank you for saying that. That that was actually literally one of my favorite episodes of season two. I wanna ask, do you leave these episodes feeling hopeful about where we are in tech governance today or or more worried about where we're going

and what comes next. I mean, my the whole theme of the podcast, if I were to zoom out, is this notion of, like, tech power in humanity. Right? Like, power technology is amazing. Tech power has shifted from strong people with sticks to to kings and queens, to politicians, to rich people, to now to people who are deploying technology. And, like, what does it mean for the humanity of all of us as that power dynamic is shifting across the world? I do leave all these conversations hopeful because I talk to people who are like, you know, what what does technically optimistic mean in the grand scheme of things? I'm optimistic because of the people that I talk to. I'm optimistic that I feel like if we can get these stories out there into the world, that more people will realize, like, they actually shouldn't be learned helplessness. Like, these are not inevitable situations. Like, you can be advocating for yourself. You can be using your wallet as purchasing power. You can do all these things. And, yes, people like us talking today need to figure out how to help everyday regular people get even more power in their voices, then I'm optimistic we can get there. And so, like, yeah, I actually do leave. I mean, I leave with more questions than I have answers in the grand scheme of things. I I leave in an optimistic frame that this is possible. We might be in the last shot. Like, this might be the we might be, like, trying to prevent escape velocity.

are, one, Instagram deploying teen teen related, accounts. And then, two, how close something like COSA is has gotten or is is getting depending on depend depending on your frame of it. And I feel both of those are from members of the public speaking out. And then people like us, people like government officials finding those needles in a haystack and enabling them to have a larger voice around it. So, you know, whether it is lots of parents sadly talking to senators about how their children have died through the harms of social media, I think forced a bunch of change on these systems. I think, like, people like Frances Haugen speaking up has forced a bunch of change and shown a lot of light in the system. Shame on us that it took so long and we had to get so far down the path before we actually can push change to actually occur. But, optimistically, I think it demonstrates that there is a path of what speaking out could actually do, what organizing could actually do in this world, what public pressure, what the press can actually do when it comes to the tech companies and the figure and helping them see the wrongs of their ways. I think there's also a place for us, you know, if we were thinking about it through a capitalistic lens, to actually go and build things that also change the public pressure. Like, I think that's firmly where someone like a Mozilla fits in or someone like a Signal Foundation fits in. Like, they're giving us actual real alternatives. And so, like, the role of philanthropy could be to help start and fund some of those before the market can actually really kick in on it. Like, actually do risk risk based capital to actually get them up up up and running and get them out the door. Because, you know, is signal the right implementation for all for every single messaging app? Probably not. But does the existence of signal force a bunch of people to contend what they're up to? Yes. Is Firefox the best example? No. I mean, like and clearly Firefox Firefox is not a majority web browser, but the existence of Firefox forces web companies to think about the world differently. And so when we come back to public interest AI, when we come back to public interest technology, that's a real concrete thing we can do, like trying to figure out how to set up real good examples, like what New York State is up to around AI, what theoretically the National AI Resource should be doing could provide these examples that then provide like, provide gravitational pull away from what tech companies are trying to portray as inevitable and just give another version of it. So I am optimistic. I I do I think it's hard. I don't think we get this for free. Like, I think it takes people like you, me, smart other people to really be pushing for all of us in order for this to happen, but I think it's possible. I love that vision. How does it tie to what you're doing at Emerson? So at Emerson, I think a lot about how these emerging technologies have a tendency to divide our society and not put them together. And it's not a commentary on the commercial sector. I'm I'm clearly of the commercial sector. One day, I probably go back to the commercial sector. But it's more just in a commentary and the incentive system that we're all operating within Right. Especially commercially. And so I think about the lens of if we think these these technologies are gonna divide society and not put them together, Emerson is just a unique organization that does philanthropy, venture capital, etcetera. We have a window to get in early on technology companies, early with technologists to try to influence them and try to bend that curve. So in a decade, it's not actually a philanthropic problem or it's a smaller philanthropic problem. So I think a lot about, like, what are early levers we can do as technologists are coming up through the educational system, coming up through the academy, as startup or as startup people and entrepreneurs are creating some of these foundational companies of the future, what can we be doing to talk to them at that point so we can try to deflect that curve in the future. It's so interesting to hear you say that. So we've been thinking a lot about that at CDT as well, including

typically, investors are gonna ask you about this and, like, the flexibility that you have. I I actually think it's even earlier than that. I think that, like, yes, changing the incentives is hard, but can we even, like, step one or two steps back? Like, as you're creating a team, the team, the people who are gonna go tackle, create your company, etcetera, can we create a diverse team? Can we create a team with a whole bunch of different perspectives around the table so we don't have a monoculture going on inside a company that's singularly focused, but one that can actually question each other and try to find the best solution for themselves or their lived experiences. I think, like, those are the types of levers that we can be doing. Commercial the commercial market is gonna be a very hard one for someone like us to go and fix. And we need help from governmental actors, etcetera, to put different incentives on the scale. So it's not just the bottom line that you're gonna report every quarter kind of thing, but, like, just even creating the team of people who are gonna work on it, I think, could be a really big deal.

Are we getting a season three?

So we're actually in we're in the I don't know what the stage to call it right now. We're in the brainstorming stage. So, yeah, we will be working on season three. The question is what form does season three take? You know, we've been debating, like, do we go to YouTube shorts? Do we go to TikTok stories? You won't see me on a TikTok story. I'll someone else send it. I was feeling very impressed. I'll I'll I'll be behind the scenes on that one. But, yes, you know, we've been bantering a different ideas around of what the theme of the season could be. But this optimistic future one, Alex, might be the one that that's the most popular one on the team right now. So that might be the one that we land on. We are here for it. We are here for all the content that you're putting out and all the good work that's coming from the Emerson Collective. Thank you so much for spending time with us today. This was really fun. This was so much fun. Yeah. Thank you for having me. This is fantastic. And for all our listeners, please feel free to join us again. I I am Jamal Magby, and thank you to Alex and Raffy for your time. Check out Raffy's podcast, Technically Optimistic, on the Emerson Collective's website and places podcasts are found. To stay up to date on all of CDT's work, please visit us at cdt.org

and follow us on Facebook, Mastodon, LinkedIn, and x at SendDemTech. And thank you all for talking tech.