Latest / Computer Says Maybe / The Toxic Relationship Between AI & Journalism w/ Nic Dawes

Transcript

The Toxic Relationship Between AI & Journalism w/ Nic Dawes · Computer Says Maybe

• Alix: [00:00:00] Hey there. Welcome to Computer Says maybe this is your host, Alix Dunn. And in this episode we are digging into a gigantic topic that I'm actually surprised we haven't covered yet on the show, but we're gonna be digging into the economics and politics. Of AI and journalism. It's a huge topic, I think, both from the perspective of a news consumer like myself. Alix: I think a lot about how AI is changing, how I find information, quality of that information, and also just the underlying community of people who work in journalism to go find information and bring it to me, the news consumer, how those people's lives and careers have changed dramatically in the last 10 years, and also how this space is gonna look in 10 years time from now. Alix: It's pretty obvious that the business models of a lot of our best journalistic outlets are bad at best and now co-opted by oligarchs at worst. We have on the show again, Nic Dawes, who was on the show a couple months ago talking [00:01:00] about how Grok is basically an attempt by Musk to reanimate his childhood apartheid days and make himself feel at home at the top of a white supremacist hierarchy. Alix: So we had him on the show for that. We'll link to that in the show notes. It is really worth a lesson if you didn't hear it first time around. But we've got Nic on the show today to really dig into the economics and politics of AI and how it's reshaping journalism. When we recorded this conversation, he was the executive editor of the city, which is a super cool journalism outlet in New York. Alix: Um, that actually maybe in some ways helped bring down Eric Adams campaign. So if you haven't looked at the city before, do check it out. It's really cool publication, but he's no longer there as of. Two or three weeks ago. So this conversation is maybe a, a portal back in time when that was his job. But he has been working in journalism for decades and has a tremendous amount of knowledge and wisdom about how to think about the way that technology might disrupt what comes next because he's been, uh, through it as [00:02:00] technology has disrupted things that have come before. Alix: Let's dig into it with Nic Dawes Nic: I'm Nic Dawes. I'm the executive editor of the city. We're a nonprofit newsroom that serves the people of New York. There's a whole history here that is a little bit forgotten about reforming journalism and reform journalism. Someone like Jacob Reese, who wrote about contaminated. Water and cholera and whose pamphlet journalism about tenement housing and public health sparked the creation of upstate reservoirs and produced the result that New York City now has some of the best order in the world. Nic: There is labor journalism, you know, that all ran all the way through the early 20th century that was about producing responsive and effective government. And it, it revealed failures and corruption and it revealed the [00:03:00] workings of the Tammany Hall machine. It was integrated with a vision of a functional civic environment, and some of its practitioners were pretty engaged in the project of building a more responsive local, state civic culture. Nic: And they didn't see themselves as a class apart in a way that I think the quote unquote professionalization of journalism, especially after the Second World War, produced this kind of elite managerial class who view themselves as a. Secular priesthood of information through whom information passes unaltered. Nic: That professional class version of we do accountability. We skillfully and carefully reveal a bad thing, and then we leave it up to others to act has really broken down in this information and political environment, and so it almost starts to feel like the default setting of journalism is libertarianism, which is not its history at all. Alix: That's so interesting. So what would they be doing alongside the uncovering of. [00:04:00] Malfeasance to create that kind of environment. Nic: I mean, some of them specifically worked at publications that grew out of movements and out of organizing. You know, in some cases, that was labor. In the case of someone like Reese, it was feeding into and sparking civic activism around housing conditions and water and sanitation. Nic: Some of them were politically involved and there were obviously risks and complexities around capture and. Patronage and all those kinds of things that went along with journalism in this mode. But there were also huge potentialities and I think trying to figure out the role that we play now does mean actually reactivating some of those potentialities. Alix: So do you wanna say a little bit about what the city does? 'cause as you're describing this, I'm like, Hmm, maybe you conceptualize the city as trying to be. Different. And I see some of the coverage as not just critiquing or uncovering, but like trying to create some type of social fabric connected to the information you're putting out there. Alix: So do you wanna describe how you conceptualize the city as publication, maybe doing some of [00:05:00] that work? Nic: The city was founded initially in response to a pretty classical sense of the crisis in local news. The New York Times had slashed its Metro Bureau in half. In 2016, the daily news had gone from 300 odd people in the newsroom to fewer than 30. Nic: The post was becoming evermore, a right wing shell. There was just less capacity. So the first thought was, we need capacity. We need people to be covering city hall. We need people in the neighborhoods to regain some of this lost. Coverage, lost sensing network, lost accountability. But very quickly it became clear that this was about more than recapturing the good old days, and in some ways that the good old days weren't all good. Nic: The city's early successes were very often around choosing journalism that could have an impact. Designing its investigative projects for impact and thinking about its relationship with New York City [00:06:00] communities in a way that. Reoriented the newsroom's position relative to readers, communities, individuals. Nic: So we really do two things. We do investigative reporting, pretty robust investigative reporting that is designed for impact, and that can be about official misconduct and corruption, or it can be about larger structural issues. So we've been at the forefront of reporting on campaign finance violations by. Nic: The mayor of New York City, Eric Adams, digging into the ways in which he's assembled fraudulent. Donation records or his campaign has assembled fraudulent donation records more structurally. For example, we've looked at construction site accident records stating back 40 years and correlated them to temperature increases, and looked at the rate of increase in job site injuries when the temperatures above 85 degrees when it's above 90 degrees. Nic: New York State has no standards. For workplace safety [00:07:00] in situations of increasing heat, the governor of New York then incorporated our reporting in her state of the state report. The Department of Labor produced its first guidelines on job site conditions in extreme heat, citing our reporting so we can both do the more individual. Nic: Classical corruption style, investigative reporting, and look at larger structural issues. So that's one piece of what we do. The other piece, and this relates very much to how we conceive accountability and civic engagement, I hope more richly is to do service reporting, civic service reporting. That means really producing deeply reported journalism that helps people, practically helps 'em to navigate complex issues, which could be about voting or your rights when you're arrested, or could be about how to get a fire hydrant legally turned on so you can pull yourself down in the street as heat theme emerging here. Nic: Or how do you join your local community board? How do you [00:08:00] access rental assistance? And our theory of the case is that if we can show up for you with advice that turns out to be accurate and relevant, then we can build a relationship with you that when our investigative reporting lands, it lands in a context of some kind of pre-existing relationship of trust and that we can think about in looking into questions of how to get things done. Nic: We can also see how things are working and we can. Show where government services are available, where they work, where maybe some of the pinch points or broken things are, and offer people meaningful resources about how to navigate, what's difficult about that. So that enables us to engage with both the effectiveness and the problems of. Nic: Local government in ways that aren't just about, uh, calling out individual points of failure. Alix: So interesting. 'cause you're making me realize that one of the ways you can do that second part of work is because you're so [00:09:00] place-based. And I feel like even if you take another publication coming out of New York, the New York Times, because they're not. Alix: Place-based. That kind of superficial, I can't remember exactly how you said it, but like the people, a part of the situation that they're covering that makes, I imagine it extremely difficult for them to connect with their readers in the same way that you do. And I feel like. Maybe this is obvious, but to me what you're describing is also part of the huge issues with the crisis of the collapse of local journalism is that increasing the information emerging from places about places isn't actually embedded in those places in a way that would allow you to. Alix: Engage with populations and actually communicate with them in ways that are like truly local. Nic: The death of local journalism is usually described in terms of the collapse of advertising revenue, the decay of print subscription models, and [00:10:00] the hollowing out of longstanding monopoly or duopoly, local news outlets by hedge funds and and listed chains. Nic: And of course all that is true, but. Rebuilding. It doesn't mean just. Refilling those same containers that we had before, especially in the context of the kind of information disorder that we live in right now. So we are trying to rebuild it along new lines and with a very deliberate intention of thickening, local civic connective tissue of enabling. Nic: People to manifest their relationship to the place that they live in more richly and fully participate in it, and to act as an enabler of them. We recognize that we aren't the convening place of the whole city, but we think we can play a role in rebuilding some of that civic connective tissue. So to give you an example, we will often convene people for a conversation about. Nic: Issues that are affecting them in their neighborhood. [00:11:00] We might, for example, go to a local library branch and hold an off the record conversation. We get a few community organizations to bring people together. We listen to them and try to produce journalism that's responsive to their needs. We take a lot of queries online as well, but we can also run the process in the other direction. Nic: So we ran a very extensive series of investigations into property speculators who use quirks in inheritance law. To rip off black families who own homes in gentrifying areas. So if four or five kids have inherited a brownstone in Bed-Stuy or Bushwick, one of them lives in North Carolina, the speculator goes to the kid in North Carolina and says, or the inherited in North Carolina and says, I'll give you $20,000 for your share, and then you don't have to worry about this anymore. Nic: You're not benefiting. Anyway. Then they use that share. To sue the rest of the family to take possession of the property and to sell the property at a very cheap price. It's a widespread and really devastating [00:12:00] practice in the areas that are gentrifying fastest in New York. So we identified this incredibly cynical. Nic: Ring of speculators who had done this in multiple instances across Brooklyn. That reporting helped to spark some legal reforms to make it more difficult to manipulate inheritance in this way. But we also held a town hall actually outta a bank branch. In Bed-Stuy with community members to hear from them what they were seeing, and also to partner with local nonprofits who provide legal counsel and do activism in this space. Nic: So that we weren't just leaving it at, here's a bad thing, or here's a bad thing and a legislator listened to us, but actually here are some resources you can use to vindicate your rights and push back. And so closing the loop of community conversation in those ways around our investigative [00:13:00] reporting allows us to show up in a very different way and to have a very. Nic: Different kind of relationship with our audience than a traditional outlet would have. I feel like Alix: this is a good segue to start talking about the role. Some people talk about AI possibly playing. So one example I hear a lot is, oh, because there isn't an outlet in X Town, no journalist is going to the city council meetings and tracking what's happening. Alix: So we could take the recording of that city council meeting, take. The transcripts and then AI can summarize and produce some reporting, I hate to use that word when describing that as a process so that there's some visibility into the machinery of local government. But like even hearing the way you're describing the value add of a local publication where it's. Alix: Two-way conversation with the people that live there and it's deeper understanding of [00:14:00] the issues because you understand the dynamics of the city, which then allows you to ask totally different questions than someone who didn't know anything about the city. I am presuming you are skeptical of the value, also the frame you used of a pre-existing relationship of trust. Alix: AI is nothing generative. AI is nothing if not untrustworthy. And I feel like, yeah, so thoughts on seeing AI as, I don't see anybody describing it as a panacea, but I do see people describing it as kind of a bandaid in the decline of local journalism. Thoughts on AI filling the gap. Nic: I think for small newsrooms like ours, and there are many that are much smaller than ours. Nic: The idea that you can have this mechanical friend who will enable you to power up your small team, not replace them, just power them up like a gunda that will take your team of five or 20 and turn them force, multiply them into a hundred. Is incredibly appealing and I think [00:15:00] that on its face, your first reaction is gonna be sure, like no journalist is gonna lose their job. Nic: We are thoughtful and careful about what we put out into the world so we could use this assistance. I think that there may be cases in which there are some things that you can do that genuinely are helpful, but I think we are currently rushing into some choices that are likely to be very, very risky, and we are thinking about them in a way that is too limited. Nic: If, for example, we we're using some kind of machine transcription service to cover council meeting or a community board meeting. And a human read it and looked for things, that would be one layer of efficiency. If the machine was producing a summary and pulling out the key points, that might be another. Nic: And if it was pulling out the key points and writing the story, that would be a third still. And I think the. Canonical answer that you're gonna get right now is, well, of course the summary is super [00:16:00] useful to our journalists. 'cause then they can drill in on the key point and one person can review the transcripts from 10 council meetings and choose which ones to cover. Nic: You can go a certain way with that argument, but we would never, we would never risk just pushing out a, you know, an automated story into the world. I think what that assessment misses is well. Whose voices were actually captured in the summary. You know, what was the operative logic of that model? When it was choosing what to highlight? Nic: What was missed, who was missed? What were the power dynamics or other dynamics embedded in those choices? Which doesn't mean you can't make them. I think you have to be really alert to the ethical value, political, and other logics that are embedded in that operation Before you go there. Agree. Alix: It's also so transactional or it's like, it's like a way of thinking about journalism as an information delivery service [00:17:00] rather than as a deeper social exercise in understanding. Alix: Helping a community make meaning, taking that example of local council meeting, like that's sonography of people in power and essentially helping them shape a narrative or a collective understanding of what's happening. I mean, to your point about like what gets missed, who gets highlighted, et cetera. Alix: It's like scaling the voice of the powerful. On rails. Do you think that having more information about what the powerful are saying is better than having overall less information about what's happening in a place? Because I feel like the idea would be that you would do good reporting and talk to people that aren't often talked to about these issues. Alix: Better understand the kind of deeper context. Nic: I think that there've been plenty of experiments in like, let's just make sure the record is out there. People don't particularly engage. [00:18:00] With the record, New York City has a setting aside, you know, more complicated things, a vast amount of open data, just vast, and you don't have to be particularly savvy to access it and process it. Nic: It helps if you can clean it up a little bit, but really where we've been very effective with public open data is when we have. Help people to make meaning out of it. And where we have put it in a context of both thoughtful investigative reporting and community conversation. So for example, we looked at the practice at many schools of calling 9 1 1 when a kid as young as six or seven is having a psychiatric. Nic: Crisis and the cops arrive and kids, usually black boys, end up in handcuffs because of a psychiatric crisis. So we had to correlate 9 1 1 database details, schools information, and we built a map of where these things are happening, income levels, what resources are available to schools. And we convened a conversation at the Brooklyn Public Library with mental health advocates, [00:19:00] parents, some kids who'd been affected. Nic: Put those facts into the world in a very different way, and ultimately, again, had some legislative and community impact. Now, we could have probably had some machine assistance in, you know, accessing that data and collating it, but the real journalistic work was the making of meaning around it and the forging of community conversation. Nic: Around it. The data was there, right? So what your institutional or social role is in activating that information, making sense out of it, and. Deploying it in a more kind of complicated democratic architecture. We need to start thinking about journalism more in those ways than as simply a kind of crude model of transparency, I would say. Alix: Yeah. Like making meaning rather than uncovering information, especially in an environment where there's been this tipping point, which Eric Sgio, I don't know if you've followed his work, but he talks a lot about moving from the age of information where there's so much value in [00:20:00] uncovering information that is not. Alix: Available and information is of highest value into an age of noise, where essentially there's so much information that, like how you add value in those spaces is completely different. And it feels like journalism continues to try and be like, here's one fact that you didn't know before. Um, or Here's a little bit more information rather than reconstituting how it conceptualizes itself. Alix: Making meaning for a community that reads it. I dunno. This also feels connected to the conversation about what objectivity in journalism is that making meaning implies that you have a perspective, which, yeah, I don't know. What do you think about that whole debate? Nic: That debate frustrates me a lot because I think that it is built around a series of pretty weak conceptions of what objectivity is, of what fairness is, and of what our role is. Nic: And I think it's a debate that really grows out of the. Transition of journalism from a blue collar craft to a white collar, elite [00:21:00] profession, which needs to have a normative framework that feels like law or engineering or medicine. And I just don't think it tells us that much. I would rather have a different set of conversations about how we do rigorous fact finding with high standards of verification, how we engage with the world and our audiences and communities, and. Nic: What mechanisms or theories we have about trust and accuracy and you know, undergraduate analytical philosophy model of objectivity. It's just not gonna give us those answers. So we waste a huge amount of time in that debate, I think. Alix: Agree. Well, let's move on to another debate. I feel like you have thoughts on, I dunno, there's two layers here. Alix: One is that AI companies are brazenly. Scraping the internet, including websites and content that are marked as saying, please don't scrape me. And basically just did that. And then we're like, [00:22:00] eventually the law might catch up to us, but we're of such a size that fines are, you know, trivial. That has had a huge effect on how media outlets, I think have seen what their opportunities are in this moment. Alix: 'cause they basically got all their content. Jacked. Um, some publications have been entering into formal relationships with these companies, but also there's this recent history of the last like 15 years. It seems like media companies have been like bait and switched tons of times by these big tech companies in ways that have been financially devastating. Alix: Do you wanna talk a little bit about what you see in the economic relationship between these big tech companies and media companies? And then maybe we could talk a little bit about. What you'd like to see in terms of how these negotiations are happening in these spaces? Nic: I mean, there's no question that media companies have been the feedstock for big tech companies for a pretty long time. Nic: And because we made such terrible [00:23:00] mistakes in the nineties and early two thousands around our response to changes in audience behavior to the move from, you know, the open web to Platformization. To the advertising monopoly at Google and Meta, we're kind of in a defensive crouch all the time, and we have gone from being Luddites to being. Nic: Tech solutionists on the back foot, low rent tech solutionists, and we beat up on ourselves essentially. You know, because of our failures early in the digital transition are price takers constantly in relation to the tech companies and platforms in particular. And we're seeing that repeated as we move into the ai. Nic: Which doesn't mean we shouldn't be engaging with them or using some of these tools, but we are not thinking clearly enough at all about the terms of the bargain. So I think that we are currently contemplating two options. One is to [00:24:00] sue, other than New York Times, um, media organization of enough institutional capacity and power, cultural and political power to push back legally on a copyright basis. Nic: Or to do a deal and get some form of intellectual property compensation. That, of course, leaves a lot of people out. So smaller outlets like the city neither have the heft to sue, nor the leverage to negotiate. So we are kind of an excluded term in that set of choices. Those of us who also are concerned about the overall. Nic: Health of the information commons also have a complicated kind of ethical dilemma to respond to, and that is, if AI powered search, for example, comes to dominate or agentic or other forms of. Question powered AI come to dominate the information space and all of it is trained on the worst crap on the [00:25:00] internet and more thoughtful, engaged journalism is not available. Nic: Then everything we do to try to build a healthier information commons happens in some less relevant, more Waldorf space, unless we can be very savvy about getting it out in other ways. So that's the trial dilemma that we face at the moment. I don't think there's a unified answer for the whole sector, but. Nic: Breaking out of this false choice between suing and doing a deal, I think is part of the task of the next two weeks. Alix: Yeah. But I think these court cases on fair use are gonna obviously have some implications that I'm kind of surprised. I haven't actually read deeply into the rulings, but it seems like there's some judges buying the argument that an AI model training on training data is. Alix: Akin to a person reading things, learning things, and then doing things with that knowledge, which seems to be a complete misreading of what the technology's actually doing. Nic: I think in the case of news, the notion that [00:26:00] what Google and AI mode, for example, produces or what GPT produces in response to a question about news, the notion that that is not some kind of derivative work just confounds my understanding of the law completely. Alix: So taking this, let's say in the next five years, especially seeing how quickly people are taking up engagement with. Let's say GPT is an example as a parasocial relationship with a knowledgeable counterpart, which seems to be what's happening. I imagine there will be an attempt by companies, if not explicit, but implicit just in the way that they're launching these products, and then people's expectation of these products changes. Alix: Even if it wasn't intentional that eventually some people, some proportion of the population is gonna be turning to these models for. News and information and a pathway to understanding the world that's similar to what you've described in a local journalism sense. Not just give me the [00:27:00] headlines, help me make meaning of how I should be engaging in my immediate city or surrounds. Alix: What do you think? That means, I mean, I know you there. There was this recent FANG piece on the intimacy dividend, which I thought was interesting in terms of like how to understand what people expect in these spaces and what they offer people that maybe isn't offered by. The kind of wild west of the web where you're like finding sources and trying to evaluate those sources and building, compiling an understanding of the world by accessing information from a variety of places, which I think is what the internet has been until now. Alix: But do you wanna say a little bit about what do you see happening for readers and news consumers and information consumers, and maybe some of the implications of that? Nic: Yes. I think that the behavior of, tell me about this thing. Tell me about how I should think about this thing. Tell me how I would think about this thing if I had the following concerns or the following ideological predispositions. Nic: That behavior is growing, you know, really, really quickly, of [00:28:00] course, and I think it's very appealing, particularly in view of some of the historical failures of the media and also in view of. The disorder of the information environment to imagine that, that you can construct this safer. Space in your conversation with an LLM, then you can have in a either a public facing Reddit type of environment where you might be exposed or a Twitter or blue sky type of environment where you might be exposed to criticism or harassment or negativity in the worst instance, and so you can make this safer sphere in which you can more fully manifest yourself in your experience of that information. Nic: It's much more of a risk to our ability to build solidarity, to create functioning civic environments, to imagine the other, to retreat into that kind of. Parasocial relationship [00:29:00] with a private platform than it is to think about ways to better manage the challenges of a disordered information environment, uh, a complex set of contending voices, noise, et cetera. Nic: So where are you as a user gonna take your trust? How are you going to make choices about what sorts of information, intermediaries or partners. Friends, you let yourself have. I think it's very concerning to imagine that, that, lemme just say fake promise of intimacy should supersede the social sphere. I mean, it's a kind of a Margaret Thatcher. Nic: There's no such thing as society. I would feel better in a kind of therapeutic and ameliorative relationship with a digital mirror than I am about my relationship with the social world, frankly. I find that pretty terrifying. And. I don't think that the role of journalism is to be in a relatively slightly more reliable feedstock for that process.[00:30:00] Alix: I love this metaphor of a feedstock. It's so good because it's, um, because I mean, if you imagine, I mean, I love the idea of of it being fake intimacy because it is. We had Dya Sidharth and Serena Agnew on from the Collective Intelligence Project, and they shared polling, which I hadn't seen or heard before, but basically the trust people have in the company that runs the. Alix: Spots are, are like the, I don't even know, what do we call these systems now? Taking open AI versus chat, GBT people don't trust the company, but they trust the product. That belies either a misunderstanding of the fact that the product isn't a walled off experience, but that actually the. Things that they're saying to these systems are being used to train these systems, that there's very little protection of the experiences they're having. Alix: They are talking to OpenAI when they talk to chat GPT, but they're, it's not conceptualized in that way intentionally. So, you know, I mean, it's not an accident that the user experie is very similar to messaging a [00:31:00] friend. Like I think that's a very intentional set of decisions if you continue. To make really good journalism and uncover really interesting things. Alix: And these models were to get faster at scraping and stealing work, or entered into formal relationships with certain publications. One, they get to decide which publications they are centering. So like who's to say Sam Altman and Rupert Murdoch aren't somewhere on an island? Talking about a partnership between two of them. Alix: Like we, we don't know how they conceptualize public interest journalism and, and what that means to them, but also. It's a, seems like a shit financial deal based on past ways that these things have evolved. It ends up being not worth it, but also like how does a user know when it's engaging? What information is being sourced from the shitty parts of the internet and what information is being sourced from legitimate ones? Alix: Like what is, what do you think that even. Looks like, Nic: I mean, they will tell you that they're surfacing links and they're getting better at showing you their [00:32:00] train of thought. That's true on some platforms to some extent, but I think what is obscured in a lot of these interfaces, and by a lot of these design choices is the naturalness and coherence of the interaction. Nic: Has an incredibly powerful, I think, nudging effect and maybe even a trust generating effect, which would explain the fact that people trust chat, GPT and not open AI because their experience of it looks like a conversation with a friend. It looks like it has a naturalness and it looks like it's sourced, so you don't actually have to interrogate the source 'cause it looks like it's sourced and feels sourced, and it gives the appearance of legitimacy in a way. Nic: Classical fake news, which dresses up propaganda under a non-existent newspaper headline and puts a photo in a caption and makes it feel like a news story relies on a similar accruer, but similar conception of how you create these effects of [00:33:00] authority or effects of relation that people are likely to rely on. Nic: Let's just take us red that that is going to happen and going to continue. But I think for public interest media, we have to find ways to interrupt and work around it. Find ways to be part of the civic conversation that circumvent. Are at a disintermediation by these tools. We may allow them to train on our data even, but we need to build different kinds of relationships with our audiences, different channels that reach them directly, that engage them. Nic: In a one-to-one relationship with us, which could be as simple as subscribing to an email or as high touch as coming to an event. We even send out physical postcards to people. We did a bunch of reporting on. The minimum heat that landlords have to maintain in New York City apartments and some tools for how you ensure that your apartment isn't too [00:34:00] cold in the winter. Nic: This is obviously usually in lower income neighborhoods, so we pulled all of the data from the city's 3 1 1 portal complaints about heat. We sent physical postcards specifically to the buildings with the most heat related complaints with a QR code on them that people could come to our story. We encouraged them to tell us what they were facing. Nic: People gave us very direct feedback. I used your tips and it worked for me. Or, here's another question that I have. That sounds like a very simple, linear example, and in some ways it is, but I think we have to find ways to work around or interrupt a relationship that is just a circuit between a user and a and a model. Alix: I'm also hearing, I mean, from. The earlier part of our conversation about how these tools might be used. Do you think it's possible to use these technologies while also trying to disrupt them? Like is there a political statement about being an organization that doesn't engage with these models, or do you think it'll over time rot the brain of publications if they're integrating these things? Alix: I dunno, is it like a Trojan horse kind of thing? You like let these models [00:35:00] in and then it's hard to continue to fight them? Nic: I think they're in, it's more important that we have. Rarely clear and thoughtful frameworks for how we engage with them than it is to be purely abolitionist. So the places in which. Nic: AI inserts itself into our workflow and our audience relationships, and pretty much everything we do are very diverse, and they are not necessarily the obvious ones. So Otta, what an incredible transcription tool. I'm an investigative reporter who now uses auto to transcribe my voluminous notes. Super helpful. Nic: I have no time and now they all sit on auto servers. It may be highly confidential. They may be subpoena able. What processes and rules and frameworks do we have for thinking about safety in a situation like that? And a reporter may just implement their own auto account without asking 'cause they think it's a cool new thing and they haven't thought through all of the safety implications.[00:36:00] Nic: Or our content management system may enable us to take an article that has been written and edited in a completely classical way. And auto generate three or four different headlines for testing to see which one generates the best result. And of course, you will have the opportunity to sign off on those headlines before they're published. Nic: So there'll be a human in the loop, but we may not have thought through the behavioral dynamics of humans in the loop and the likelihood that we will accept those recommendations and the concern that we might have outsourced the process of ethical deliberation that we have in relation to a headline to, uh. Nic: Platform that has no investment in our traditions of. Ethical thinking about those things. Or we might do some vibe coding and you know, much more quickly be able to spin up something that's of real value to our readers when we can't afford an engineer. So we need to be able to think about all these pieces of our production and value chain, what the ethics are, what the economic implications are, what values [00:37:00] we're embedding, um, and how they speak to our larger value framework. Nic: I would say that. Our ability to generate information as a public good and deploy it into a functioning civic context has been. Critically weakened by platform economics. Now we are being asked not only to hand over the most valuable things that we produce, not just the information, but the relationships with communities and readers that we have. Nic: We're being asked not only to hand those things over to platform companies, but become so fundamentally their customers. To outsource to them so many of our core functions and processes that we really just become a small cog in their machinery while being even further economically weakened. I think circling the drain is, is, is almost too polite a way to describe the depth of the extractive process [00:38:00] that we are, that we are subject to and the threat to our. Nic: Public function that now exists. I, I think there's no question of that. At the same time, I think that. We can't pretend that we live in digital spaces. We function in a world that is completely penetrated by these systems. So they are here and I think we're more likely to be exposed to more risk if we pretend that they're not. Nic: So I think we need to draw pretty clear, pragmatic, but values-based internal lines about. Where they implicate different pieces of our production chain and our relationship with community. So it's not about so much definitely navigating and living with complexity as it is recognizing that we need strong frameworks to live with these systems because we just do and we need coherent strategies [00:39:00] to bypass them and offer. Nic: Alternatives where we can, and we are tiny and fragile and held together with duct tape. So those are just also realities that we have to reckon with. Alix: Thank you so much. This is super interesting to talk about the journalistic work you're doing in New York. Nic: Thanks, Alex. I think it's very useful to talk about this in the same context as. Nic: Thinking about outsourcing government functions, thinking about where civil society intersects with these systems, and I think those of us who are working on the journalism piece of it could use more of that kind of reflection. So it's been good to talk some of it through with you Alix: next week. I sit down with Sam Gregory, who's the Executive Director of Witness, which is a super interesting organization that's focused on all things human rights and video. Alix: And I wanted to sit down with Sam for a long time, but with the SOA two announcement where OpenAI basically in a very [00:40:00] cavalier way, launched a synthetic video factory that's of higher and higher quality. And we get into this with Sam, but also just like is set up to be abused by abusers. I really wanted to get Sam's take, and more than that, more than just his take on what's happening now, which he is the person that I go to when thinking about the role of video and synthetic media in our modern information environment. Alix: I also just wanted to talk to him about how. All of this has evolved over the last 15, 20, 25 years because he's basically anticipated a lot of this stuff. He's not alone. There's a community of people that have been working on the same set of problems for a really long time and have basically anticipated this exact moment of Sora two for a long time and have tried to encourage. Alix: Companies, civil society, governments, philanthropy, to prepare for this moment. And largely that hasn't happened, but he's just been at the forefront of understanding and anticipating all of these things. And so we, we take a look back at the sort of the way that questions about audio, visual content, and human rights and [00:41:00] information environments has evolved since the founding of Witness. Alix: So that is next week's show with Sam Gregory. Also, we are at MossFest. When this episode goes live, I'll be in Barcelona with the team. My colleagues Marin and Jake will be with me live. And uh, if you are around, let us know. Also if you are around and listening to this, and MozFest hasn't ended yet, you should take a look at the program. Alix: 'cause we are gonna be involved in several sessions on the Friday and the Saturday and the Sunday. So do take a look at that. Some of those will be live streamed. So if you're feeling the fomo. And you're not gonna be in Barcelona. Um, we will be broadcasting a couple of conversations we're hosting live. Alix: And with that, thanks to Georgia Iacovou and Sarah Myles for producing this episode. See ya.