> ## Content Index > Fetch the complete content index at: https://werd.io/llms.txt > Use this file to discover other available public pages before exploring further. # Notable links: September 23, 2026 - URL: https://werd.io/notable-links-september-23-2026/ - Published: 2026-10-23T13:09:44.000Z - Updated: 2026-10-23T13:09:44.000Z - Description: Two reports on AI in news; how it can intermediate relationships; and some final thoughts about Foursquare. - Author: Ben Werdmuller - Tags: Link roundups *Most Fridays, I share a handful of pieces that caught my eye at the intersection of technology, media, and society.* *Did someone forward this to you?* [*Subscribe for free*](https://werd.io/notable-links-october-16-2026/#/portal)*.* --- I sat down with New\_ Public to talk about building in the open, local news, the open social web, and more. [I enjoyed the conversation very much; here it is](https://newpublic.substack.com/p/why-this-platform-is-for-everyone). --- ### [The News + AI Field Report](https://cronkite.asu.edu/knightcenter/field-report-AI-in-news) The [News + AI Field Report for 2022-2025](https://cronkite.asu.edu/knightcenter/field-report-AI-in-news), produced by [Djordje Padejski](https://www.djordje.padejski.net/) in conjunction with [ASU’s Knight Center for the Future of News](https://knightcenter.asu.edu/), is a landmark analysis of how newsrooms are experimenting with AI. The [interactive app](https://aimap.futureof.news/) is also a really useful way to explore the 160 projects covered by this research. I’ve been mulling over one of the findings described in the summary: > “Newsrooms are most worried about the business threat from AI and have built the least for it. Interviews reveal widespread anxiety about the effects of AI on site traffic, audience behavior and news business models. Yet relatively few projects focus on revenue, subscriptions or fundraising.” 20% of projects were business-related, but every fundraising project and over half of the business projects were grantees in the American Journalism Project’s AI programs. If those newsrooms are stripped out, the numbers drop significantly. I think there’s a subtext worth talking about: I suspect the innovation projects listed here (that weren’t a part of a specific cohort) are a temperature check on how newsrooms are building technology / product experiments in general. If that’s the case, it implies that news is sticking to its comfort zone and spending much more time innovating on editorial concerns, and much less time experimenting with solutions for its existential business threats. Reader-facing chatbots are a different angle on this: I’d call them a way for newsrooms to say they’re experimenting with AI without actually addressing a core user need. Newsroom websites are absolutely not where the chat layer should live (except, perhaps, for the use cases that provide deeper archive search for subscribers): AI users already have their own chatbots, which are the interfaces for *all* their AI-powered information. There are deeper ways AI could address user needs; the trick is that they would involve experimenting with the fundamental ways newsrooms do business. In turn, that requires both a willingness to do so, and much deeper collaboration between teams to make it happen. So then the question becomes: how do we not just beef up journalism’s experiments in its business models, but the excitement and appetite to do so? Often the product teams that might be the owners of these kinds of experiments don’t (quite reasonably) own business models; how can those teams collaborate effectively? Funding from AJP clearly made a difference — is outside scaffolding required for this work to take place? How might product teams empower every team in the newsroom to run experiments? These are much bigger questions than AI – and they’re questions newsrooms probably should be grappling with. In the meantime, I’m very grateful this analysis exists, and I hope it’s renewed to capture the year that follows. --- ### [The Three-Layer Cake: Public-Interest Information in an AI-Intermediated Age](https://www.journalismfutures.org/three-layer-cake) Gina Chua and Adiel Kaplan at the Tow-Knight Center for Journalism Futures have put together a report summarizing a year of convenings and conversations about how journalism might react to its intermediation by AI. I was privileged to take part and to read a draft version of this report. I love the framing here: the three-layer cake is digestible (so to speak), and I think the conclusions are sound. The worrisome thing for journalism is that most newsrooms live in the *creation* layer — the creators who supply the information which informs answers and summaries — which is also the part of the stack where participants have the least leverage. So we need open, shared, common infrastructure. I think we also need strong portability standards. I’d argue that another word for the *consumption* layer as described here is *profiling*; tools at the consumption layer need to know a lot about their users so they can provide the most relevant information. As the report says: > “This is the layer that holds the relationship with the end user, where their data resides and the understanding of their information needs and desires accumulates and deepens over time. Owning that relationship will be one of the prizes of the AI age; not only will it unlock deep knowledge of individuals, it could lock them into particular providers.” And any solution needs to keep this in mind: > “At the consumption layer, \[points of leverage\] involve working to ensure that the knowledge of people’s wants and needs is not the monopoly of organizations that may not have the public’s interest at heart.” Profile data is accumulated, private, and usually non-portable. The consumption layer also represents the greatest risk to creators from platform business decisions, both in terms of the business models of those platforms and which creators they choose to feature. Open portability standards and protocols — combined with full user control of their data — can scale, create useful competition, and empower users in the process. --- ### [Team Dynamics Weaken When Employees Turn to AI Instead of Colleagues](https://hbr.org/2026/10/team-dynamics-weaken-when-employees-turn-to-ai-instead-of-colleagues) Relationships take time, effort, and practice to keep alive. So in a sense, it shouldn’t be a surprise to see this finding from a survey of more than a thousand desk workers across 20 industries: > “The more they turned to AI instead of a teammate, the more their trust in each other and their ability to coordinate seemed to erode.” It turns out that if you start asking an AI system instead of your colleague, coordination and trust drop by at least 20%. These technologies are still relatively new; I imagine that over time this will continue to drop. People who feel isolated in their work today will feel more so as this trend solidifies. Similarly, people who feel untrusted today will only feel more so as their teams continue to ask AI systems instead of them. That doesn’t mean AI is always inherently bad. Like all work processes, it demands that we think hard about culture: what kinds of behaviors and feelings do we want to promote? What’s important to the organization? And do we care more about an individual’s productivity than creating a high-performing team? (We shouldn’t.) This also rings true: > “People consistently underestimate how enjoyable a human conversation turns out to be. Whether the motivation is fear of judgment or a desire not to impose, avoiding a colleague is hardly a new workplace behavior. AI now provides an alluring alternative that forgoes the social and emotional costs of getting support.” One way we can counteract this is by making asking for help (and giving it) an explicit norm in our team cultures, and leading by example by doing it ourselves. In remote or hybrid teams that rely on tools like Slack, that also means doing it in a non-private channel, so that other team members can see it and it can become normalized. Although a bit of a cliché and genuinely heinous if they’re done poorly, I think standups can also be really helpful here, if you can do them live. Here, everyone goes round a circle and talks briefly about what they’re working on, with explicit needs for help made as part of the update. Implicit requests for help can be answered once everyone has gone around the circle. Can AI help? Totally. As the authors discuss in this piece, you can use technology to help conversations to happen asynchronously, particularly on teams that spread across multiple timezones, and it can actively facilitate connections between people. Let’s do that, instead of displacing our people and our relationships. AI can be a facilitator between people, but it should never be an intermediator: our relationships, our communities, and our companies will suffer. --- ### [Some final words on Foursquare](https://teendrama.substack.com/p/some-final-words-on-foursquare) I freaking loved Foursquare. It was a wonderful way to explore cities with the help of my friends: in the same way that Upcoming was my best way of figuring out which events I should be going to, Foursquare let me know which restaurants, bars, parks, etc, were worth checking out. If Chris ranked a cocktail bar, you know it was amazing. Aaron always found the best coffee spots. And so on: my friends were better recommenders than any other source I had. My own map lists every state and country I’ve been to over the last few decades. If I’m back in Amsterdam, I can figure out which Indonesian food place I really loved, because there it is on my map. (It’s [Paon Bali](https://paonbali.nl/), by the way. You’re welcome.) So this was sad to see as an outcome. I think there are a bunch of important lessons here – not least the dangers of raising too much too soon without tethering the business to revenue. But at the same time, Foursquare launched during a startup explosion, and this kind of fundraising was the norm; I don’t fault Dennis for his decisions. The bigger problem seems to have been the private equity leadership more recently. I can’t think of a time when private equity *hasn’t* led to toxic leadership and a strip-mined company with an inequitable deal for employees. Which leads me to an obvious problem. That check-in data is intimate stuff. It probably shouldn’t be something private equity can just inherit and do *whatever* with. Journalists and activists should hopefully not have been using Foursquare to check in; if they did, that data can almost certainly have been bought for analysis. And then a future-facing question: how might we have these kinds of social effects *without* sacrificing privacy? Regardless, I’m pouring one out for Foursquare. I loved it. Thanks to everyone involved for building something that I found genuinely meaningful.