Renate Nyborg

The through-line

Nyborg's fixed point is loneliness, and everything else she says radiates from it. She describes it as "the greatest battle we fight in our lifetime in addition to the climate crisis," an existential problem on the order of the climate crisis, because "if we stop having close friends being in relationships having children etc etc that is an existential threat to who we are" [1]. That framing explains an apparent contradiction in her career: she ran Tinder, and now predicts the slow death of dating apps [1][2]. Her account is that two decades of work on social connection, across Apple, Headspace and Tinder, coincided with loneliness rising rather than falling, and she now sees "the generation of people that built some of those apps moving out to build apps or products that help you return to real life connection" [1].

The second constant is that the answer is not abstinence from technology. "The solution isn't everyone throw your phone out the window like that shift is happened" [1]. She wants to use cheaper, more accessible AI "to make technology good for people again," while insisting the technology teach skills rather than substitute for the relationships themselves [1].

On the slow death of dating apps

Her prediction is "that uh we will see the slow death of dating apps and even Mass social media as we know it and a triumphant return to people meeting through community and forming close social connections that way" [1][2]. She does not expect them to vanish: "I think things like dating apps they may become like uh eBay they'll probably always be around in some formal shape," but the swing towards meeting online is reversing in some countries [1].

Her evidence is generational. She has consistently worked on products popular with eighteen-year-olds, and argues "this is where we always need to look in terms of the future if we're going to predict what's going to happen next" [1]. Gen Z increasingly meets people through friendship groups and dates them afterwards: a Pew study showing 50% of Gen Z women were friends with a partner first, and a Canadian study suggesting as much as two thirds of eighteen to twenty year olds at college became friends before dating [1]. That, she notes, is "pretty much the opposite of meeting someone on a dating app uh that you never would have come into contact before" [1]. Her explanation for why swiping worked in the first place is that it was a permission mechanism for a moment when meeting strangers online was still unusual: "I swipe right on you you swipe right on me it's okay for the stranger to talk to me now like I've invited you into my online personal space" [1]. As online and offline blur, that mechanism loses its purpose, and people meet wherever they already are, on Instagram, on Animal Crossing, on fortnite [1].

She is also blunt that the industry misreads its own competitive landscape: "the real competition as much as the media likes to make it out this way was never really between Tinder versus hinge versus Bumble like if someone's actively online dating they're typically going to be on multiple of these apps at the same time" [1]. The real shift was towards meeting people anywhere online, which reads as a closer proxy to real life because the person is already connected to your friends [1].

On what a relationship AI should and should not be

Nyborg names AI companions as a live risk: people "forgoing real connections real relationships real friendships um in favor of AI Companions," a risk she believes should be counteracted with something more ethically designed [1]. Her metaphor for Meeno is culinary rather than romantic: "we're more like rat Remy from Ratatouille um so not a not a fake girlfriend or boyfriend but instead of helping you make amazing soup we're teaching you relationship skills" [1]. A collaborator she describes as an AI luminary, Andrew, shares that view [1]. The other recurring image is "taking a flight simulator before you know going off into the the jetplane r of real social life," a place to rehearse rather than a destination [1].

The scope is deliberately wider than dating: friendship, romance and family all count, and the product is meant to help people understand their attachment style and how to form deep personal relationships [1]. Her argument for why software is needed at all is that the relevant knowledge is stranded: it is "trapped in books in tons of podcast CS in the minds of therapists and most importantly in our own minds in our own experience," and "at the moment there is no lexicon for really what it takes to communicate openly" [1]. Supplying that lexicon, she argues, gives people "more like of an operating system to all of the ways in which we'll meet people both online and offline" [1].

On designing the product

Generative AI removes the need to march people through a course. Because the models are trained on how people actually write and speak, Meeno can "learn from the way that people want to talk" instead of forcing users down a prescriptive path built for the technology's convenience [1]. In practice that means a two-minute onboarding that captures who you are, what you care about and one goal or issue, followed by something closer to a Choose Your Own Adventure, with music, animations and nods to games like Zelda so the experience feels entertaining [1].

The core interaction is empathetic questioning that helps a user reflect on a specific situation, then rehearse it. She cites a beta tester, Jimmy, preparing a difficult conversation within his friendship group and learning to say plainly that he was nervous about it reaching Snapchat, which she treats as a case study in how much of connection is simply having the language and then the confidence to ask for what you need [1]. What separates Meeno from Google or ChatGPT, in her account, is recall: over time it builds a map, surfaces connections and flags patterns, up to observations as granular as "hey this seems to be something that you're always struggling with on a Tuesday" [1]. The output is deliberately constrained: "advice uh and suggestions critically not Solutions though we've heard up all of our test users they don't want to be told what to do but they'd like to have a bit of a menu and then decide for themselves" [1].

On whether AI weakens the skills it helps with

She treats atrophy as "one of the most important questions we can be asking ourselves," and answers it with the calculator: "we didn't ask people to keep doing everything in their own head we made uh the requirements for math harder," raising the bar because the tool existed [1]. Applied to schools, where there are calls to block generative AI in subjects like maths and English, she prefers inverting the classroom: a student drafts at home with ChatGPT, and the classroom work becomes improving and critiquing it [1]. On the anxieties about outsourced sincerity, the ChatGPT wedding speech, the machine-written thank-you letter to parents, her test is counterfactual: if the tool is the difference between someone never writing to their parents at all and writing for the first time, then starting a conversation and building on it, "that's better than just having no starting point at all" [1]. Her testers back this up in small ways, one practising a text to a mother he had been annoyed with and discovering that taking the first step changed her behaviour [1].

On why Gen Z, and the appeal of not being judged

Meeno is focused on Gen Z partly because early studies showed markedly more interest there, and partly because the need is sharpest: 83% of US Gen Z consider themselves lonely sometimes or all the time [1]. She points out the inversion this creates. People over fifty are "much less lonely ironically than jensi that's because they got their friends from when people weren't looking at their phones all the time," and she is explicit that the product is not built for them [1].

What draws young users, she says, is the absence of judgment, "in particular for the 18 to 25 year olds because there is this fear of being judged there is this fear of being canceled for saying the wrong thing" [1]. The value is a place to think out loud where there is no human on the other side to react emotionally [1]. She hears the same worry repeatedly from people who grew up online: say something and "is going to be everywhere straight away if it gets into the wrong hands" [1]. Her stated ambition for eighteen to twenty-five year olds matches their own: "their biggest dream is to meet someone quite and quite organically" [1].

On why software alone will not fix loneliness

Nyborg does not claim her product is the answer. Part of the erosion of relationships and close friendships, she argues, is "a decline in public spaces and things like that," which no app addresses [1]. She positions Meeno as holding one piece, social skills education, that "needs to go hand in hand with how people meet in public spaces um people feeling uh like they're in a safe environment where they can meet people safely for the first time" [1]. She has connected with loneliness researchers for insight on partnering, and frames the work as requiring innovators, policymakers and society together: "it's going to take a village" and "it's going to take a lot of us to come together to try to tackle this problem" [1].

Takeaways

• Predicts "the slow death of dating apps and even Mass social media as we know it," with dating apps surviving as niche infrastructure, "like uh eBay" [1][2].
• Reads the future from eighteen-year-olds: Pew found 50% of Gen Z women were friends with a partner before dating, and a Canadian study put it as high as two thirds among eighteen to twenty year olds at college [1].
• Argues the swipe was a permission mechanism for an era when meeting online was unusual, and loses its function as online and offline blur [1].
• Rejects the industry's own framing of competition between Tinder, hinge and Bumble, since active daters use several at once [1].
• Positions Meeno against AI companions: "not a not a fake girlfriend or boyfriend but instead of helping you make amazing soup we're teaching you relationship skills," a flight simulator for real social life [1].
• Builds for recall and pattern-spotting over time, and gives users "a bit of a menu" rather than solutions, because testers do not want to be told what to do [1].
• Answers the atrophy objection with the calculator: raise the bar for what people are asked to do rather than ban the tool [1].
• Targets Gen Z because 83% of US Gen Z report loneliness and because fear of judgment and of being cancelled makes a non-human listener appealing [1].