[Chad Davis] 12:02:20 Hey everybody, I'm Chad Davis. I'm the Chief Innovation Strategist at Nebraska Public Media. Welcome to our August Public Media Innovators webinar. Introducing myself and Amber Samdahl from PBS Wisconsin. Yeah, thanks for joining us. Also, thanks as always to [Chad Davis] 12:02:39 NETA, who does a great job helping facilitate and promote these webinars as part of our public media innovators peer learning community. If you are inclined to do so, just pop into the chat and say hi and tell us where you're. [Chad Davis] 12:02:54 where you're joining from today, but I am glad you're joining. [Chad Davis] 12:02:58 This is gonna be a really, I think, important webinar. I ran across New Public's work, probably it was in June, I think. It started basically like showing up everywhere in my LinkedIn feed, and… [Chad Davis] 12:03:14 I was familiar with some of the work they had done before, because they are kind of tied in with the concept of the filter bubble, which some of you may know from the early teens. And I was really interested to see that they now had a new take on what the future. [Chad Davis] 12:03:33 of, sort of, online community and interaction could be looking like in an era of AI. Um, so when I checked it out, I was like, we have to… we have to get this in front of our public media innovators crowd. So, uh, thank you for. [Chad Davis] 12:03:48 for being a part of this today, and I would like to introduce Angela Quicksey. She is the Managing Director of New Public, and you know what? I'm gonna be done talking now and say, Angela, go ahead, just take it away. Folks, put questions in the chat. We're gonna have a good long run for Q&A at the end, and I think as you watch this, you're gonna. [Chad Davis] 12:04:07 You're going to have lots of things you're going to want to kind of click to follow up on. So Angelica, go ahead. [Angelica Quicksey] 12:04:14 Hi, y'all. I am Angelica Quicksey. I'm a managing director here at New Public. And I'm based in Seattle. Our team is actually remote and distributed. [Angelica Quicksey] 12:04:24 I co-authored the After the Feed report with our founder and executive director, Eli Pariser. So I'm really excited to speak to that today about AI and social technology. We're calling this talk AI and the Future of Community. [Angelica Quicksey] 12:04:40 And I actually joined New Public to stand up a project with public broadcasters in North America and Europe. So I speak a little bit to that at the end and some of our additional subsequent work. So if you don't know New Public, we are a nonprofit product studio dedicated to healthier community and connection online. [Angelica Quicksey] 12:05:01 A lot of work doing R&D, building new tools, and then doing research around what works and what doesn't for allowing people to engage in healthier ways that bring people together instead of atomize them and create hostility online. [Angelica Quicksey] 12:05:18 So I'm going to go ahead and get started. [Angelica Quicksey] 12:05:21 Bringing us back to what Chad described, which is the filter bubble. [Angelica Quicksey] 12:05:26 our co-founder, Eli, gave a talk in 2011 introducing the idea of the filter bubble, which became a book of the same name. And the thesis at the time was that. [Angelica Quicksey] 12:05:39 algorithms were increasingly personalizing what people see. Um, not tuned directly to them, but to people that look like them, to their interests, to the way that they engage and show up online. And that, if left unchecked. [Angelica Quicksey] 12:05:54 these personalized algorithms would have negative consequences for people and for democracy. [Angelica Quicksey] 12:05:59 And I think we've seen that a lot of the predictions that Eli made in that original talk and book have played out in really negative ways. [Angelica Quicksey] 12:06:09 15 years later, we are at another inflection point. So as generative AI is surging, it is being distributed in how people work and in the actual infrastructure of how it's built and trained and operated. [Angelica Quicksey] 12:06:27 We're at this sort of moment where the social technology landscape is changing, it's shifting, and it's shifting very quickly. So we started this research project in the winter, so end of last year, and published our report in May. [Angelica Quicksey] 12:06:42 Uh, to ask what trends will most significantly shape how humans connect to one another, build relationships, have experiences in online spaces in the next two years. So we were looking ahead to 2028. [Angelica Quicksey] 12:06:55 And rather than being a repeat of the last era, how could these trends actually support human thriving and connection? [Angelica Quicksey] 12:08:46 So getting into the technology and what this actually looks like, I want to really emphasize that generative AI and what we're going to call the Agentic interface, and I'll speak more to that in a little bit. [Angelica Quicksey] 12:09:00 It's more personal and persuasive than anything that came before it. So we took a step back and looked at the trends of how people are engaging with technology and asked what this looks like. So right now, it's this idea that you wake up, you pull out your phone, you create your AI agent. [Angelica Quicksey] 12:09:17 And you ask, you know, what do I need to know this morning? [Angelica Quicksey] 12:09:21 And this new interface can respond with everything you need to know, curated for you, for your interests, for your day, and in the format that you prefer. And between May, when we released this report, and today, we've already seen that Google. [Angelica Quicksey] 12:09:38 Spotify, Meta, and others have released versions of this new feature and of this interface. [Angelica Quicksey] 12:09:45 And what's important is that it doesn't look the same for everyone. It is hyper-personalized. So for your very online friend, the morning update might look and feel a little different. For busy parents like myself, I dropped my one-year-old off at daycare this morning. [Angelica Quicksey] 12:10:01 the update comes while you're on the move. So it isn't even necessarily the same interface, the same phone or screen that we're used to, but might be voiced, but might come in another format. [Angelica Quicksey] 12:10:13 And these Agentic interfaces are pulling from traditional media sources, but also synthesizing information that's specific to you, your chats, your calendar, and it's curating that information for your particular day. [Angelica Quicksey] 12:10:30 We're calling this the era of the Agentic Interface. We're calling it the Agentic Interface and not just, sort of, chat or… [Angelica Quicksey] 12:10:37 You know, generative AI, because while chat is the dominant interface in the summer of 2026, we don't actually know what the final form will be. This stuff is moving so fast, even from when we started this project to when we released it to today. [Angelica Quicksey] 12:10:53 We've seen so many changes in model capabilities, in how people are engaging with this technology, and I think that it's going to continue to shift, and we'll see new forms over the next. [Angelica Quicksey] 12:11:05 Six months, 12 months, 18 months, two, three years. [Angelica Quicksey] 12:11:09 I want to be clear that in this new era, the social feed is no longer the default interface, but the feed doesn't necessarily go away entirely. We're not saying that social feeds aren't going to exist anymore. [Angelica Quicksey] 12:11:24 People are still gonna scroll. That's where parasocial relationships and entertainment will continue to thrive. [Angelica Quicksey] 12:11:30 But Agentic interfaces will be the new intermediaries for information about the world around you. [Angelica Quicksey] 12:11:37 And these are interfaces that are built for an audience of one. So it could be a filter bubble of one where it is increasingly personalized and becomes the nexus through which you access information and connection. [Angelica Quicksey] 12:11:53 There's a few trends that we identified in our original report that I want to dive into. [Angelica Quicksey] 12:12:00 Uh, around, uh, how people are engaging with information, how they're accessing it, and what's happening to the dominant social platforms today. [Angelica Quicksey] 12:12:09 So the first is this idea that we're seeing a changing of the guard, a set of new gatekeepers controlling what the public sees. [Angelica Quicksey] 12:12:18 The old gatekeepers were institutions. This was the curatorial power of newspaper editors, of TV news anchors, who focused attention on particular topics, you know, on what was on the front page of a local newspaper. [Angelica Quicksey] 12:12:35 And that shifted in the filter bubble era that Eli identified. So we started to see platforms and algorithms be the dominant sort of curatorial layer shaping what people saw and what the public would engage with. [Angelica Quicksey] 12:12:50 And now we're entering this era of a new gatekeeper, a new, new gatekeeper, where we have LLM-powered interfaces that are tuned to you specifically. [Angelica Quicksey] 12:13:00 At the same time that they're becoming this gatekeeper, they're fracturing the incumbents. [Angelica Quicksey] 12:13:07 So while the social platforms are still a place where people go for entertainment. [Angelica Quicksey] 12:13:11 They're becoming a lot less social, they're becoming less safe. This is the golden era of scams everywhere, but particularly online. [Angelica Quicksey] 12:13:22 People are — a lot of the traffic that we saw on the internet in the previous era was actually reciprocal. It was people asking questions and getting answers, and that is becoming eroded. People are increasingly going to AI to ask questions rather than to strangers on the internet. [Angelica Quicksey] 12:13:39 And that reciprocity actually has an impact for how people engage, who they trust, where they get information. [Angelica Quicksey] 12:13:46 And at the same time, all you have to do is open your LinkedIn feed to see a lot of low-quality AI-generated slop. [Angelica Quicksey] 12:13:57 And over the last couple of months, just earlier this year, we saw that automated accounts and content created by agents or bots has surpassed human content online. So in a very short period of time, we're seeing the amount of human content that's posted on the internet. [Angelica Quicksey] 12:14:15 is being outnumbered by content that's created by agents or bots. [Angelica Quicksey] 12:14:21 I want to spend more time on trend number three because I think it's particularly relevant for public media. [Angelica Quicksey] 12:14:31 as AI is changing how much content can be generated, what content can be generated, where it can be posted, the speed and scale at which people create information on the internet, it means that. [Angelica Quicksey] 12:14:45 the, uh… [Angelica Quicksey] 12:14:46 Currency of the previous era, likes, views, follower counts are increasingly gameable. So where before the internet rewarded whoever could capture the most attention by those metrics, those metrics are very unreliable now. [Angelica Quicksey] 12:15:02 And this is accelerating a trend that was already happening, where the attention economy is now being supplanted, replaced by the trust economy. [Angelica Quicksey] 12:15:14 We're seeing that both from users who want to find people, creators, friends and family that they trust, but also creators and media organizations themselves are sort of leaning into trusted relationships. [Angelica Quicksey] 12:15:30 Because that's where, their audiences are going. [Angelica Quicksey] 12:15:35 So if trust is the new economy. [Angelica Quicksey] 12:15:39 Rather than those likes and follower counts, we're seeing this emergence of what we're calling thick reputation. This is reputational measures that are contextual or slow or built over time. [Angelica Quicksey] 12:15:53 So rather than anchoring on a follower count of 10,000 followers or a verified badge, you're starting to see this sort of interest in people that have contributed thoughtfully to a community for two years or vouched for by people that I trust. [Angelica Quicksey] 12:16:10 So it adds nuance, context, sort of thick reputation to the traditional metrics that we saw online, and I think this is an interesting design challenge for the social technologies of the future. [Angelica Quicksey] 12:16:24 And we're seeing that this is really where people are going. [Angelica Quicksey] 12:16:29 Where the open internet is filling with synthetic engagement, people have already been and are increasingly migrating their real social lives to these smaller, more private spaces. [Angelica Quicksey] 12:16:39 Group chat are where people are engaging. That's where I'm spending more of my time than on the open internet. [Angelica Quicksey] 12:16:47 Maggie Appleton calls this the cozy web, and it's happening at every scale, whether it's about the neighborhood. There was this article that came out a few months ago about the group chats that changed America. [Angelica Quicksey] 12:17:02 Business and political decisions are getting made as well, which means they're getting made quietly and private, out of the view of the public. [Angelica Quicksey] 12:17:12 I'll get to the last two trends and then speak to some of our learnings about our work with public media and where this is going. [Angelica Quicksey] 12:17:21 So, the fourth trend has even shifted since we initially published this report, where we believe that we're moving into this era of software abundance. The exciting side of this is that more non-technical folks can build things for themselves and their communities. [Angelica Quicksey] 12:17:40 We hear about Vibe Coding, Claude Code was released early last year, and of course, AI-based coding tools have exploded. [Angelica Quicksey] 12:17:50 And sort of anecdotally, I've seen people building new high trust micro communities for their book clubs, for their running groups. And that's really exciting that product builders don't have to write code. [Angelica Quicksey] 12:18:05 But at the same time, we're seeing a lot of bad software. And I do think this is gonna shake out over time, but, uh, just in the last… I think the New York Times published, uh, maybe a month ago since we last put out this report, uh, this… [Angelica Quicksey] 12:18:20 news that submissions to Apple's App Store have surged dramatically. They're higher than they've been since 2019, but the number of user downloads hasn't budged. So there's this widening gap between the new apps, the new software that's flooding the market, and actual user value. [Angelica Quicksey] 12:18:37 And then lastly, we did an analysis of the functions that social media plays in people's lives and looked at which ones are sort of at risk of being replaced by agents. And for the most part, the functions that. [Angelica Quicksey] 12:18:53 were around connection and belonging were really resistant to being replaced by AI. [Angelica Quicksey] 12:19:00 And I think that's really important because… [Angelica Quicksey] 12:19:06 are already being supplanted by John. [Angelica Quicksey] 12:19:14 a certain institution tends to be more durable. Um, and this is true even when you take into account. [Angelica Quicksey] 12:19:21 um, AI use cases like, uh, companionship, um, and companion agents, uh, companion-oriented agents, where, uh, people are… [Angelica Quicksey] 12:19:31 coming to AI for companionship, but in general, people will only use AI if it's better than the best available human. So even in those use cases, it's usually filling a need that isn't being met in their current life, and they'll turn to humans if. [Angelica Quicksey] 12:19:47 and took to the people around them if that becomes available. [Angelica Quicksey] 12:19:52 So how does this relate to public media? [Angelica Quicksey] 12:19:55 I think it goes without saying that AI will drive and is driving massive changes to how we communicate online. [Angelica Quicksey] 12:20:05 But right now, despite these changes, most Agentic interfaces are actually designed for human to agent interaction. They're not interjecting as much into how humans communicate with humans into collaboration tools. [Angelica Quicksey] 12:20:19 They're really emphasizing providing information or taking action on a human's behalf. [Angelica Quicksey] 12:20:26 So that means that, um… [Angelica Quicksey] 12:20:29 Media organizations and any organization or institution or product or business that produces information, anything that can be summarizable, should be planning for what Xu Weifang calls a B to A to C world. [Angelica Quicksey] 12:20:50 that's whatever the Agentic interface of the future is, uh, is gonna be the intermediary between the information that a business or an institution produces and where humans consume it. [Angelica Quicksey] 12:21:03 Um, and so just that understanding how agents or generative AI systems intake information and translate that for humans is a really important design challenge to wrap our heads around. [Angelica Quicksey] 12:21:15 But she also says that human communities still need local relevance and cultural nuance. The defensible moat lies in understanding specific communities' realities and facilitating shared understanding. [Angelica Quicksey] 12:21:29 And I think that's an opportunity for public media and local media in particular, because while you provide information, you're also designing for a world of human-to-human relationships. [Angelica Quicksey] 12:21:43 And that means that the direct relationship with your audience is more important than ever. [Angelica Quicksey] 12:21:48 So here's a little bit about what we've learned about human-to-human connection online from our work with public broadcasters. [Angelica Quicksey] 12:21:55 First is that people often make sense of information in relationship with others. [Angelica Quicksey] 12:22:02 While individuals' relationships with institutions are in decline, particularly in the U.S. [Angelica Quicksey] 12:22:08 Their relationships with individual journalists, whether they're associated with traditional media or they're independent journalists, are still holding steady. [Angelica Quicksey] 12:22:17 And that's where people are increasingly going for news. They're also going to people, individuals that they trust, whether that's online or IRL. And that means that conversations with neighbors, friends, coworkers are part of where people make sense of the world. [Angelica Quicksey] 12:22:34 And so we're seeing institutions like KQED, I believe they spoke about their discord to this group before, lean into creating communities where people can engage with one another and make sense of the. [Angelica Quicksey] 12:22:50 the shows that they're producing because that's part of the informational environment and the informational landscape. [Angelica Quicksey] 12:22:59 Second. [Angelica Quicksey] 12:23:01 How people engage online varies more by age or generation than by language or nationality or location. So we did research into the motivations for participating in public conversation across public media. [Angelica Quicksey] 12:23:18 of four public media audiences in Canada, Belgium, Switzerland, and Germany. [Angelica Quicksey] 12:23:23 And I personally went to Toronto, Montreal, Brussels. [Angelica Quicksey] 12:23:29 And to Mainz, Germany, and then my colleagues went to Switzerland to do focus groups, workshops, one-on-one conversations with audience members from these communities, both in English and in their languages. [Angelica Quicksey] 12:23:44 And before we started this project, our partners insisted that their audiences were unique, the way they engage with content was completely different, and somebody who German versus High German in Germany, they have nothing in common. And so we would need to do very specific audience research in each place to understand how their audience is engaged in public conversation. [Angelica Quicksey] 12:24:08 And that wasn't true. Yes, there were differences. The audience in Germany tended to be more skeptical of technology than the audience in Canada, for example. But actually, young people across Canada and Belgium and Germany and Switzerland had so much more in common. [Angelica Quicksey] 12:24:27 than the older audiences in each of these places. And that was where we actually started to play with is the generational differences and then the type of person who wants to lean in online versus the wallflower or the person who tends to observe online. [Angelica Quicksey] 12:24:44 So I thought that was really an interesting insight that I wanted to bring to this group. [Angelica Quicksey] 12:24:48 Though, of course, there was a lot of room for error across language and culture. Sometimes our translations didn't quite land, and that created some… [Angelica Quicksey] 12:24:58 comedy throughout, but I just was really struck by the similarities across language and nationality and the differences across generations. [Angelica Quicksey] 12:25:10 And then lastly, people are really hungry for locally rooted information, and I think that's important for this group. We've partnered with the Civic Information Needs Census, who are doing longitudinal surveys. [Angelica Quicksey] 12:25:24 every year asking the same questions to audiences throughout the country. [Angelica Quicksey] 12:25:30 And they're finding that people find it is easier to get information about what's happening around the US or around the world than what's happening down the block in their neighborhood, in their city or town. [Angelica Quicksey] 12:25:43 And so far, that is something that AI can't provide. [Angelica Quicksey] 12:25:48 So that's actually why we launched Roundabout. And I'll do a quick plug. We created this platform and launched it in January in five pilot communities around the US. It is a platform that brings people together. [Angelica Quicksey] 12:26:04 neighborhoods or groups of neighborhoods of about 10 to 50,000 people and allows them to share information about what's happening in their community. There's buy-nothing groups, there's events. And importantly, we've built a moderation stack and approach that is rooted in local stewards who know their neighborhood and know the people in these spaces. [Angelica Quicksey] 12:26:27 So we've had zero instances of toxicity across the platform and across these five Copilot communities, now six because we just launched in Southeast Portland last month. [Angelica Quicksey] 12:26:37 And we, this week, announced that we're partnering with Chicago Public Media to bring the platform to all of Chicago and some of the surrounding areas, because as we've been building this platform, which is rooted in community. [Angelica Quicksey] 12:26:53 We've also been engaging with public media in North Carolina and now Chicago because they know their neighborhoods, the type of information that they need, and are looking for different places to engage with their audiences and to distribute their content. [Angelica Quicksey] 12:27:11 And we have felt like this is a really rich area for partnership to be thinking about content and community together in one place. [Angelica Quicksey] 12:27:21 So with that, I will end. But I am available — I'm excited to answer any questions that you all have. Ask me anything. Here are places to stay in touch. And I've actually linked out the full After the Feed report. [Angelica Quicksey] 12:27:37 and her research into motivations and archetypes for public conversation at the end here, as well as some additional information about, uh, our Chicago Public Media Partnership and, um, Shuwei's work on the B2A2C Design Challenge. [Chad Davis] 12:27:53 Thank you, Angelica. Yeah, questions in the chat folks, but I have a couple to kick us off and maybe Amber will as well. I'm just gonna kind of work backwards a little bit through what you were kind of talking about. The roundup I think is fascinating to me. [Chad Davis] 12:28:09 The timing of the announcement couldn't have been better for us. And just doing the webinar because it's like, hey, you're here. [Chad Davis] 12:28:18 how does that, like… I mean, I know it's not like a Nextdoor, and Nextdoor… people have very strong opinions about that platform, but how does it, like… how does your vision for it, or how is it functioning thus far, how does that kind of compare to, like, a Nextdoor, which a lot of us kind of at least have a basic understanding of? [Angelica Quicksey] 12:28:39 Yeah, we very early on thought of it kind of as an anti-next door, actually. And we had looked at that platform and how they've designed it quite a bit in the early days of creating Roundabout. [Angelica Quicksey] 12:28:54 So actually, before we started Roundabout, we set out to do an R&D project around local communities and information with the explicit goal to not build a new platform. We said, we want to meet people where they are. [Angelica Quicksey] 12:29:09 We want to experiment with getting people information on existing social platforms. What can we do to make these healthier and to create opportunities to find locally rooted and contextual information and content there? And we just found that the existing platforms are —. [Angelica Quicksey] 12:29:27 kind of designed against those purposes. Everything that we did ran into the limitations of moderation on a Facebook, or the sort of design and accessibility of a Discord. [Angelica Quicksey] 12:29:41 And that was what led us to decide to create Roundabout. [Angelica Quicksey] 12:29:46 There's a couple things that differ — make Roundabout different from any other social platform. One is this idea of stewards or stewardship. So we're designing — spending just as much time designing for what you would think of as the mods or the moderators. [Angelica Quicksey] 12:30:01 as we are for the end user, because we are selecting a steward or steward pair for every community that is playing a curatorial role in shaping the space. So it's not entirely algorithmic. There's a homepage that you see that has been curated by somebody who is based in this neighborhood. [Angelica Quicksey] 12:30:21 knows the space, knows what's important, and is highlighting content for people, and is also part of developing the rules around moderation and inviting people into the space. And so where, you know, a Facebook or Facebook groups. [Angelica Quicksey] 12:30:35 One of our co-executive directors spent seven years at Facebook working on Facebook groups, and they always sort of dismissed or spent a lot less time thinking about that role than about. [Angelica Quicksey] 12:30:51 increasing engagement, increasing revenue. And so we really wanted to emphasize the role of stewards and stewardship. And then we have this approach to moderation that is rooted in all of the best practices around moderation that we learned from PSI and from our partners and other researchers in the trust and safety space. [Angelica Quicksey] 12:31:11 that does not only post moderation of flagging things that have been posted, but does things like, you know, nudge you before you post something that violates the community rules. And this is based on the idea that comes from gaming, that most of the bad content that ends up on the internet is. [Angelica Quicksey] 12:31:29 not from bad actors, but is from good people who are having a bad day. And that's not to say that we shouldn't be designing against threats and bad actors in these spaces. But if we design with the mindset that you can interrupt the good person who's having a bad day from posting the thing that's going to create a flame war. [Angelica Quicksey] 12:31:50 Then you design the moderation system in a different way. [Chad Davis] 12:31:54 That's actually fascinating to me because literally within the last hour, a group of us were talking about potential comment moderation around games. There's a, you know, we're looking in Nebraska at rolling out a crossword. [Chad Davis] 12:32:10 Uh, instance, uh, game, uh, that. [Chad Davis] 12:32:12 will actually be available to anyone in public media, uh, when we kind of get it out the door in a couple months. But thinking about, like, okay. [Chad Davis] 12:32:20 we really like the idea of maybe opening it up to comments. We use a platform for testing things right now called Crosshair, and the commenting has been really great. But, like, I guess I'd be curious, what are the best practices that we all might be able to take away? You kind of talked about trust and safety best practices. Are there any others you could. [Chad Davis] 12:32:38 to kind of highlight. I liked the nudge and whatnot, so. [Angelica Quicksey] 12:32:45 I don't want this to sound like a cop-out, but one of the things that we've really emphasized is choosing the practices that make sense for you and your community. And this was something that we did encounter with our PSI partners, where even the audiences across different shows engaged in different ways. [Angelica Quicksey] 12:33:06 the hard-hitting news show audience for ZDF versus the sort of lifestyle story audience for ZDF, um, like, engaged with the content and with one another in different ways, and one. [Angelica Quicksey] 12:33:18 Was a little more, um… [Angelica Quicksey] 12:33:21 spicy than the other, uh, and sort of willing to engage in conflict in different ways. And, uh, and our, our PSI audiences also have really strong. [Angelica Quicksey] 12:33:33 feelings and moderation approaches around curse words, for example, versus a different type of Reddit community where that might not be a major issue. So spending the time upfront to determine what you want to allow and whatnot has been. [Angelica Quicksey] 12:33:50 a really important part of the work and one of the things we've. [Angelica Quicksey] 12:33:55 gone back and forth on is like for an institutional community or audience. [Angelica Quicksey] 12:34:02 you want the institution, you want CBC Radio Canada to be setting those community norms, versus for something that's rooted in the community, you want the community to be involved in setting those community norms. So that's one thing that we spent a lot of time thinking about. [Angelica Quicksey] 12:34:17 And then for whatever system that we design, we design — we used to use —. [Angelica Quicksey] 12:34:26 the Perplexity API that Google Jigsaw had created in collaboration with some larger media organizations, and they've decided to deprecate that, but we… [Angelica Quicksey] 12:34:37 like, we've found that the current models, like, the current generative AI models have actually surpassed the Perplexity API, so even just designing your own. [Angelica Quicksey] 12:34:47 um, your own prompt that is sort of sourcing or scoring the content, um, before it gets posted has worked really well. Like, you can actually design something custom today more than you could two years ago. [Angelica Quicksey] 12:35:02 And then lastly, having some… putting some real thought into what hap… like, how people can… [Angelica Quicksey] 12:35:11 Flag or once you make a decision, like putting thought into what it looks like to. [Angelica Quicksey] 12:35:17 like, combat or to, come back to to. [Angelica Quicksey] 12:35:23 disagree with that decision. Like, there's a lot of work around… that we've published around juries and tribunals as part of, like, online mediation or moderation spaces, and in some cases, like, that works really well. In some cases, that's a little bit too robust for what it is that you're trying to do. [Angelica Quicksey] 12:35:40 But, um, I think there's lots of different approaches. I'd be happy to share a couple of places where folks can go. We really love the ProSocial Design Network, which actually lists all of these different approaches. [Angelica Quicksey] 12:35:54 It shares which ones are really rooted and supported by the research and which ones are not, or which are still kind of not figured out. [Angelica Quicksey] 12:36:03 There's so much in this space that it's hard to say, here's the top five best practices for how you should approach moderation. [Angelica Quicksey] 12:36:11 Yeah, I was about to share the. [Chad Davis] 12:36:13 Oh, sorry. Thought I'd save you though. [Angelica Quicksey] 12:36:14 No, no, you got — you beat me to it. [Chad Davis] 12:36:19 So, I referenced Nextdoor earlier, but it sounds like Reddit is kind of more the model, and it feels like… [Chad Davis] 12:36:26 thick trust is kind of a bit like Flare on Reddit? Do, I mean, am I kind of thinking along, am I thinking right about that? Or the idea about thick trust was kind of something that I, I spent some time initially when I saw your, your deck kind of dwelling on maybe. [Chad Davis] 12:36:41 Could talk a little bit more about that? [Angelica Quicksey] 12:36:43 Yeah, um… [Angelica Quicksey] 12:36:45 So, we borrow from how Reddit has approached moderators and moderation in their communities. I actually think that Nextdoor is still, like, a good corollary. In the early days of Nextdoor, they had these sort of… I forget what their. [Angelica Quicksey] 12:37:02 They were called if they were like neighborhood captains or what have you. But they deprioritized their role over time. So this kind of — and they didn't design explicitly for stewardship over time. So it existed, but it —. [Angelica Quicksey] 12:37:17 It wasn't really… [Angelica Quicksey] 12:37:21 prioritized or surfaced in the network. [Angelica Quicksey] 12:37:24 And you see that happen in a lot of different places because people are idiosyncratic and it's hard to maintain consistency over time when you're really relying on people as part of your platform or your moderation practice. [Angelica Quicksey] 12:37:40 And part of our research into local spaces really showed that stewards face burnout because there's, you know, so much of holding together the community, it falls on their shoulders. And so one of the things that we're trying to be thoughtful about is. [Angelica Quicksey] 12:37:56 what it looks like to reduce the burden on the stewards of a particular space. [Angelica Quicksey] 12:38:01 And then FLARE is one version of reputation management system. We actually worked with the Yale Justice Collaboratory to do a case study on different versions of reputation management systems. I've just dropped the link into the chat. [Angelica Quicksey] 12:38:18 There's lots of ways to do signaling and verification online. All of this, though, is rooted in the previous era of social media, and so I'm really interested to see… we're exploring, like, well, what does this look like going forward? How do you signal reputation? What does it look like to… [Angelica Quicksey] 12:38:37 to demonstrate trust on the internet. Because I think about this a lot. New Public is rooted in urban planning. I'm a trained urban planner. And when you go to a public space and you are people watching, you can engage and be present in a public space without being legible. [Angelica Quicksey] 12:38:55 in the way that you have to be online. Being in an online space and not participating is… [Angelica Quicksey] 12:39:05 is always read as being a lurker. And there's something, like, negative and kind of creepy about being a lurker, whereas, like, sitting on a bench in a public park is not coded as lurking. [Angelica Quicksey] 12:39:15 So how you show up online just because it's a world of zeros and ones looks different. And we're always thinking about what it takes to make that legible, how you can design around that. [Angelica Quicksey] 12:39:30 Again, as these Agentic interfaces start to… [Angelica Quicksey] 12:39:36 proliferate or continue to proliferate, but also start to engender new forms of design. [Angelica Quicksey] 12:39:42 It makes different kinds of reputation management systems possible. And I would be — I'm excited to explore this a little bit more going forward. But this paper or this case study was produced, I think, last year. So it's more rooted in traditional social media. [Amber Samdahl] 12:39:59 Thank you for also making me feel better for, uh, when I am just in a community reading, and feeling like I am being a lurker. So thank you, I'll think about that analogy with sitting on a park bench. [Amber Samdahl] 12:40:11 Um, we do have a question in the chat, um, how can AI help public media organizations advance email marketing and audience engagement? I think there's a lot of interesting points you made about the connections around trust and human-to-human connections. What are your thoughts around that? [Angelica Quicksey] 12:40:28 Alright, um… [Angelica Quicksey] 12:40:30 I am not an email marketer, so I'm sure there are a lot of people that would be able to speak to that better. [Angelica Quicksey] 12:40:36 Uh, but I will say that newsletters have been one of the most promising new features for Roundabout. Like, people… [Angelica Quicksey] 12:40:45 people are… and this is not an AI problem, this is… this has just become a thing, like, and if you're a parent and you've got a million. [Angelica Quicksey] 12:40:53 notifications from your kid's school or daycare, like, you've experienced this, that there's just so much content coming at you from everywhere. Um… [Angelica Quicksey] 12:41:02 So. [Angelica Quicksey] 12:41:02 part of what AI does really well is synthesis or summary of the information that's around you, and one of the things that we experimented with before we built Roundabout was a newsletter that sort of pulled up. [Angelica Quicksey] 12:41:17 uh, using AI, sort of automatically surfacing, like, the best things for the roundup of all the things that people were sharing in somebody's local community group. And that has… people have really liked that, like, they like. [Angelica Quicksey] 12:41:31 roundups. They like, um, getting something that's tailored to them, that's something that's in their inbox, uh, that's not. [Angelica Quicksey] 12:41:39 Yet another substack, but it's pulling from… [Angelica Quicksey] 12:41:42 a bunch of different places. I think one thing to be wary of, though, is that… [Angelica Quicksey] 12:41:49 AI can help you with email targeting, it can help you with A-B testing the right, um… [Angelica Quicksey] 12:41:56 the right subject lines. It can help you with testing with synthetic audiences. There's lots of things that AI can help you do to better target, to better engage your audience, to make content that's more engaging. And everyone else is doing the same thing. [Angelica Quicksey] 12:42:13 Um, and so my concern with thinking about AI. [Angelica Quicksey] 12:42:20 as a, purely as a tool for that kind of, like, I actually am really interested in, like, narrative. We're doing a lot of work around narrative change and how AI can be used for audience testing, message testing, for, you know, testing with synthetic audiences, um, and there are limitations to that. [Angelica Quicksey] 12:42:38 But we know that, like, you know, Adobe is doing SEO for, um, for agents now, and, like, people are increasingly thinking about not only your audience as the, um. [Angelica Quicksey] 12:42:52 as the endpoint, or as the consumer of your… the information that you're sharing, but also somebody's agent. Like, increasingly, you should be thinking about your… what you're creating as something that somebody might feed into their chatbot, or something that might get surfaced to them through this Agentic intermediary. [Angelica Quicksey] 12:43:12 Shui called it the B to A to C design challenge. But later in that article, she calls it the B to A to A to A to A to A to C design challenge because it's not just a single intermediary. You might have multiple. [Angelica Quicksey] 12:43:28 Agentic layers in between what you put out there and what somebody actually consumes. [Amber Samdahl] 12:43:34 That's really interesting. I was listening to a podcast recently, and they were joking about how their inbox was flooded with newsletters that were often somewhat written by AI, and then they had an AI agent that was filtering them down to just pull out the headlines. [Amber Samdahl] 12:43:50 the AI was sending newsletters to another AI, you know, interface that was essentially weeding them out, so… [Amber Samdahl] 12:43:57 I'll complicate it as we consider our future. [Amber Samdahl] 12:44:00 Looks like there's another question in the chat. I'll just jump in and share one more. Are there proven effective uses of AI for public media fundraising? We use it now for deep research. Any other uses in that space? [Angelica Quicksey] 12:44:15 I probably can't speak to that. I wouldn't be the right person to speak to that. I wonder, actually, if anyone in the audience has examples of AI for public PDF fundraising, since this is the. [Angelica Quicksey] 12:44:31 Public Data Innovators Group. [Angelica Quicksey] 12:44:32 And someone else asked the same question of how can AI help publicly the companies fundraise? [Angelica Quicksey] 12:44:39 The one thing I will say is some of our… [Angelica Quicksey] 12:44:44 Like, I think the jury's still a little bit out on this, but some of our… [Angelica Quicksey] 12:44:49 work has been with, like, places like QQAD, or people that are experimenting with building community, um, and… [Angelica Quicksey] 12:44:57 One of the things they're exploring is whether the community building also creates a deeper relationship with the audience that then leads to more people donating. And that's one thing that we're — I think the jury is still out. We're still exploring. [Angelica Quicksey] 12:45:13 is it something that our public media partners are experimenting with? But the public media organizations that we work with are so, um, different in where they get their money. Because we work internationally. [Angelica Quicksey] 12:45:27 And we do work with US-based public media, but with Canada, Switzerland, Germany, and Belgium, they're… they're… [Angelica Quicksey] 12:45:37 mandate where they get funds, how much they rely on audiences versus government, the extent to which they can advertise. For example, in Switzerland, they're not allowed to rely on advertising revenue as part of their mandate versus in all these other countries they can. [Angelica Quicksey] 12:45:55 It's not something that we've been able to focus on just because the monetization strategies and revenue strategies have been so different across the board. [Amber Samdahl] 12:46:05 Well, on that note, also, Anthony has a question. Uh, can you speak to monetization in the research you have done and the platforms you're building? [Angelica Quicksey] 12:46:13 Yeah, so for our work with public broadcasters, we're not the ones who are monetizing. [Angelica Quicksey] 12:46:23 I know that our work with CBC, they've really seen the PSI toolkit that we've built, which is a sort of… I've spent a lot of time speaking around about… less so about our PSI toolkit, which is a kind of commenting system, but it's anchored around topics and questions, as opposed to specific articles. [Angelica Quicksey] 12:46:41 So it's durable across different articles that are about the same topic. [Angelica Quicksey] 12:46:49 CBC and Radio Canada, which are the. [Angelica Quicksey] 12:46:53 Quebecois French and, um, English-speaking sides of the house have both used it for major topics like, um. [Angelica Quicksey] 12:47:00 The most recent space trip from the US astronauts or major topics around sports and boys in sports in Canada, which was an interesting conversation. [Angelica Quicksey] 12:47:17 They've seen that as part of their strategy of getting people to create accounts with them and to engage with their content in different ways because they have decided that you need to create an account to engage in the commenting system. And that's been part of how they continue to build that audience. [Angelica Quicksey] 12:47:32 And then bring them deeper into their different systems and audience strategies. [Angelica Quicksey] 12:47:40 For Roundabout, we are supporting it through a mix of things. It is… [Angelica Quicksey] 12:47:47 It's partially advertising-based, but not surveillance advertising. So we want to be thinking about local, contextually relevant advertising as opposed to then target advertising that's based on the data that we can extract to get from people. [Angelica Quicksey] 12:48:04 For us, what we… [Angelica Quicksey] 12:48:07 are concerned about when it comes to the existing social media platforms is the surveillance part of their advertising models, not that they do advertising at all. [Angelica Quicksey] 12:48:18 Audience contributions. And then right now, we are still philanthropically supported, and we've been partnering with local foundations to support the product. We're thinking about where it goes from here and spinning out as a public benefit corporation around the middle of next year. [Angelica Quicksey] 12:48:34 But I would be happy to share what more we can about monetization strategy as we get a little bit further. We're only six months into the product. [Amber Samdahl] 12:48:47 Thank you. Um, Nancy asks, what indicators do you use to measure the success of Roundabout, and are you including content in the space beyond the community members? [Amber Samdahl] 12:48:58 And I'm curious, you know, in six more months, I'll be curious to hear how that evolves too. [Angelica Quicksey] 12:49:04 Yeah, we have spent a lot of time thinking about measures, and we use some measures like weekly engaged users, not weekly active, but sort of thinking about how people engage that are more. [Angelica Quicksey] 12:49:19 Traditional metrics, but we're also. [Angelica Quicksey] 12:49:23 have been working with KX who is who did the. [Angelica Quicksey] 12:49:28 the Community Information Needs Census, and then our partner, Talia Stroud, at the Center for Media Engagement at the University of Texas at Austin, to think about other types of metrics that speak more to the community health. [Angelica Quicksey] 12:49:41 Let me see if I can. [Angelica Quicksey] 12:49:44 pull some of that up. Um… [Angelica Quicksey] 12:49:47 Because we're thinking about not only the traditional acquisition and activation and retention, but also, is this a welcoming space? Do people feel like their time is well spent on this platform? And we're thinking about measuring those through pulse surveys on the platform. [Angelica Quicksey] 12:50:05 And then we are trying to do something that, to our knowledge, no social platform has done, which is actually run in RCT on the people's participation on the platform, where we would do sort of an experimental design where we're testing. [Angelica Quicksey] 12:50:22 for the… we're enrolling some number of people to participate, and then looking at measures of, um, social trust, agency, and belonging are our North Star metrics here at New Public. [Angelica Quicksey] 12:50:32 Um, we rely on the belonging barometer, uh, I think that comes out of UC Berkeley, and then some other metrics around agency and around social trust. [Angelica Quicksey] 12:50:42 to understand how this platform is actually driving those things as opposed to just the traditional engagement metrics. [Angelica Quicksey] 12:50:53 And we have some ways to measure social trust and belonging. [Angelica Quicksey] 12:50:58 through surveys, but the gold standard would be a randomized control trial that we would do in collaboration with the University of Texas at Austin. [Angelica Quicksey] 12:51:07 And then the content in the space beyond the community members, we it varies from space to space. We've been. [Angelica Quicksey] 12:51:18 I think we were collaborating with NC Local in North Carolina for one of our communities in North Carolina because we're focused on Alamance County, and they have a newsletter called Alamance Fabric that is targeted at that specific community, and they've done a bunch of community listening. [Angelica Quicksey] 12:51:33 to determine what their information needs were before we even started on this project. We've been bringing in some of that content for that community, and we are planning to work with Chicago Public Media to bring in information from their properties. [Angelica Quicksey] 12:51:49 But I… we're still figuring out what that'll look like. We haven't launched in Chicago just yet. We'll be launching later this fall. [Chad Davis] 12:51:58 I have another question. I'm going to jump in here real quick. [Chad Davis] 12:52:03 The, you talked about SEO for agents, something Adobe's kind of… [Chad Davis] 12:52:09 uh, digging into, um… I've been thinking a lot about, kind of, like. [Chad Davis] 12:52:13 Again, kind of inspired by this sort of Agentic Interface era thought process that you kind of kicked off. I mean, this notion that in the future, like public service media is really about feeding, at least in part about feeding these. [Chad Davis] 12:52:30 Call bots, scrapers, you know, the AI models that become the interface that kind of feeds daily news, daily truth to people, and in a way that, like, that's the new broadcast, like, flash forward 15 years or 10 years, like, that's what, that's gonna be, like. [Chad Davis] 12:52:47 what the new broadcast is gonna be. At this point, it's 2026 still, like, I see like Time Magazine is creating, you know, a specific parallel web presence for AI. Is that something that we in public media should be? [Chad Davis] 12:53:05 thinking about? Have you guys thought about that a little bit? Any thoughts on that? [Angelica Quicksey] 12:53:12 I mean, the short answer is yes. You should be thinking about it in that way. I would say that you would think about it in two different — you would think about those as two different audiences, the human audience and the Agentic audience. That's the A and the B to A to C. [Angelica Quicksey] 12:53:28 Because the… the person, the consumer, the audience member is still gonna end up with that content, but thinking about how it's going to get translated for them, and how… [Angelica Quicksey] 12:53:39 How machines parse information, decide what is relevant, decide what is true is going to be an important part of this. So it's not that far in the future. I think that commercial entities are jumping on this faster because. [Angelica Quicksey] 12:53:55 You know, um… [Angelica Quicksey] 12:53:57 Adobe wants their information, if somebody goes to ask about Photoshop, for their canonical information about Photoshop to end up there. And retail is thinking about this because they want their rug or couch or coffee maker to be the thing that ChatGPT recommends if you're asking for something. [Angelica Quicksey] 12:54:16 someone about this. But, um, Bloomberg's head of audience is already thinking about this, about the machine as part of, um, the audience for their content, and… [Angelica Quicksey] 12:54:28 I'm blanking on who made this distinction, but… [Angelica Quicksey] 12:54:34 We're thinking about this in a couple of different ways, like one is. [Angelica Quicksey] 12:54:37 what is information that is summarizable versus non summarizable? What is information that people are going to ask from ask their agent for versus go to themselves like we're seeing a bifurcation in. [Angelica Quicksey] 12:54:52 public media and in, um, in private media of, uh, what… where… what people are going to media organizations for. Like, is it for, um. [Angelica Quicksey] 12:55:04 just the newsletter, the information, like, or is it for events? I think people are increasingly building events businesses where they can connect with and hear directly from the sources that normally, you know, the media organization would go to. You're seeing that with Semaphore, who has built this. [Angelica Quicksey] 12:55:21 events business on top of their editorial. You're seeing this with local media who are getting funding from local philanthropy to build events businesses. It's part of the revenue strategy for some people, but it's also one of the things that's durable. [Angelica Quicksey] 12:55:37 when agents or agentic interfaces are increasingly becoming the intermediary for information. So it's something we're thinking about. We're seeing different types of organizations experiment with different strategies. And I would say 100% it's something that public media should be thinking about. [Angelica Quicksey] 12:55:56 this, the machine is part of your audience. It's also the intermediary between you and your audience, and understanding how information will get translated is part of. [Angelica Quicksey] 12:56:08 part of the work of anyone producing information in 2026 and beyond. [Chad Davis] 12:56:14 Cool. [Chad Davis] 12:56:15 Well, I don't wanna hijack the rest of the webinar with that. I could go deep on that, but if anybody in the audience is interested in following up on that with me, I've been thinking more about this and I'd love to go deep on that at some point, maybe in a product capacity too. [Chad Davis] 12:56:31 Yeah. [Angelica Quicksey] 12:56:32 Yeah, it's something where I'm going to be spending a lot more time on in the next two or three months because we're bringing together some experts and philanthropic leaders who are interested in the information environment. [Angelica Quicksey] 12:56:46 Thinking about what is the natural endpoint of this? Where does this go? How do you influence the intermediaries? How do you ensure? [Angelica Quicksey] 12:56:57 I really hate this phrase, but people are really concerned about epistemic decay, and how do we know what is true and what is not, and who says what is true and what is not. That's already become. [Angelica Quicksey] 12:57:11 the post-truth era or whatever, it's already become a problem with the existing social media platforms and existing ways that people are making sense of information online. There is definitely a world where that could get worse in the Agentic Interface era, and we want to make sure that. [Angelica Quicksey] 12:57:29 We, again, don't have a repeat of the 2010s and sort of ending up in the place that… [Angelica Quicksey] 12:57:38 more dystopic version of our information environment, but what does it look like to preserve humanness online, to preserve the ability to determine what is true from what is not, the ability to make sense of the information about the world around you, and to access. [Angelica Quicksey] 12:57:55 real reporting and information about the world as well. [Angelica Quicksey] 12:57:59 So I'll have more to say about that later this year. [Chad Davis] 12:58:01 I will… [Chad Davis] 12:58:03 Very cool, very cool. Um, yes, I'm gonna have a conversation with Jim and I when I'm done about epistemic decay, um, and what that is. But it definitely feels like the AI era is, like, the revenge of the philosophy majors. It's like, we just, we held on long enough, and like, and now… [Chad Davis] 12:58:18 And now we're relevant. We have one more question from the chat, and then I think we'll kind of wrap it up. It is, how can Roundabout be kind of used to gather information from viewers, members? I think in the public media space, some of us know about Harkin. [Angelica Quicksey] 12:58:21 Yep. [Chad Davis] 12:58:36 Uh, which I think is owned by IndieGraph now, um, so, uh, shout out to IndieGraph and Harken, but I think maybe that, that this question kind of comes from that, that standpoint of, like, can we, can this become part of our, our content, uh. [Chad Davis] 12:58:52 source. [Chad Davis] 12:58:52 Sourcing. [Angelica Quicksey] 12:58:53 Yeah, the Harkin folks and the Indiegraph folks are friends of New Public, and we've thought a lot about this piece. I think what's been interesting is the way that our tools for. [Angelica Quicksey] 12:59:09 public broadcasters, the sort of commenting system that we've built, have started to go in this direction, both in the, like, Ask Me Anything, Engage with a Journalist direction, and in the. [Angelica Quicksey] 12:59:18 How can we gather information from you as well? And I know that there's actually… [Angelica Quicksey] 12:59:26 that one of our Hannes Grossinger has been building something similar that's hyperlocal and a sort of crowdsourced information in Switzerland. And so I know a lot of people are experimenting with this. [Angelica Quicksey] 12:59:40 For Roundabout, I think that the jury's still out about how this will help public media organizations gather information. We're still figuring out what this looks like and what the role of the journalist will be versus the role of the steward. [Angelica Quicksey] 12:59:55 who will be a non-journalist who's part of stewarding the community, facilitating conversation, curating information. But I'll be excited to speak to that after we launch in Chicago and we're getting more into the weeds with the Chicago Public Media team. [Angelica Quicksey] 13:00:13 I almost feel like we should have Trey and, um… [Angelica Quicksey] 13:00:18 Trey is the head of Roundabout, and Kaziana is the group product manager who's working specifically on the Chicago.com project. I feel like, you know, early next year, you should invite them back to speak about that. [Chad Davis] 13:00:29 Yeah. [Chad Davis] 13:00:31 Oh, we'd love that. I was thinking, yeah, six months down, maybe just checking in and seeing how the experiment's going and what's learning. So, yeah, that's absolutely, like, we'll get in touch offline on that. [Chad Davis] 13:00:42 Angelica, thank you. Like, this has been, like, a fascinating hour. I knew it would be, but you delivered, and so I really appreciate you making time for us, and also kind of crafting, like, this presentation for public media. You kind of took the bigger thesis, and you brought it. [Chad Davis] 13:00:57 to the audience and very much appreciate that. [Chad Davis] 13:01:02 As I let everyone go, just a quick… not a reminder, because I didn't tell you at the beginning, but our September webinar is going to be returning us to the topic of immersive media. They've got some really interesting use cases and examples, both from within public media and without. It's going to be on the 17th. [Chad Davis] 13:01:21 Of September, I think we're gonna do it at 1 Eastern, that makes it 10 Pacific, and uh, yeah, um… [Chad Davis] 13:01:28 put the link in the… in the chat there for you to click on that and get registered. So, um, Angelica, thank you again. Uh, we really appreciate the time today. So, see everybody, uh… [Angelica Quicksey] 13:01:38 Yeah, thank you. Thank you for having me. Take care. [Chad Davis] 13:01:40 See y'all next month.