Dating is broken. Can AI fix it?

I’ve developed a dating-app swiping rhythm — left, left, left, left, left, right, left, left, left, left, left — the kind of muscle memory I can do while doom-scrolling an Arsenal match or while balancing precariously on the metro. The apps promise infinite possibilities; my thumb experiences them as an infinite queue. My friends tell me I’m too picky. Maybe I am. Or maybe I just read “Pride and Prejudice” for the first time when I was 11, letting Mr. Darcy set the bar a bit too high for real-life adult men. Add in a steady diet of romance books, and I’ve ended up with a brain that can turn a grocery-store glance into a sweeping subplot and still can’t find a digital match out of a lineup of men holding up dead fish.

I’ve downloaded and deleted and redownloaded the usual suspects — Tinder, Bumble, Hinge, Raya (where, yes, I saw some celebrities; no, I can’t say who). And it just so happens I’m a journalist who writes about AI and gets bombarded with glittering promises about how the technology is going to make my life easier. If AI can draft my emails, why not see if it can help with the hardest consumer product in existence: my love life.

Could AI help me meet my soulmate — or at least a guy I’d want to go on a second date with?

​​AI dating apps right now are doing a very specific magic trick: They’re trying to take the two worst parts of modern dating — choice paralysis and first-message dread — and hand them to a model.

If that sounds dramatic, researchers have essentially tested the non-dramatic version. A 2025 paper in “Media Psychology” found that evaluating lots of profiles can degrade decision-making — the “more swipes, worse choices” hypothesis in lab-coat form. That tracks with lived experience. The queue doesn’t just exhaust you emotionally; it degrades your judgment as you keep shopping for your soulmate. And into that exhaustion walks AI, offering to do the hard parts for you.

Swiping toward my destiny

The Western astrologer my friends once paid to read my birth chart told me my Aquarius sun and Pisces Venus mean I’m “built with an interesting internal thermostat,” because I “crave independence but tend to love in watercolor.” A Tibetan astrologer told me Mars finally maturing in my chart suggests I might enter a relationship around my 32nd birthday, which is, funnily enough: today, Valentine’s Day.

Outsourcing my romantic forecast to strangers and planets and then opening an app that insists the next swipe could change my life feels like a thoroughly modern thing to do. It’s a perfect little parable for dating right now: We want fate, but we’d also like a tool that edits fate into something more A/B testable.

The legacy players in the dating game have leaned into AI with the urgency of companies that have watched growth slow and churn tick up. But they’re not creating an experience so much as patching up the old one, like a broken-down building getting a new lobby.

​​Tinder is stuffing the swipe era into an AI exoskeleton: Chemistry for “better” matches, Photo Selector to comb your camera roll using on-device biometrics, and Game Game — an OpenAI-powered tool where you can practice flirting, like it’s a foreign language you forgot in middle school. Bumble is shipping generative bios and replies, talking up AI photo-picking, and adding a “report AI images” button — because nothing says romance like a synthetic-human help desk. Match Group says Hinge’s AI Core Discovery Algorithm lifted matches and contact exchanges 15% since March; it has added Prompt Feedback to rehab your onboarding. Grindr, meanwhile, is building an AI-heavy premium tier (see: $499.99 in some early pilots) that reads like SEO for flirtation.

The pitch is simple: Reduce friction, and make the funnel flow again. All of the product design tells you what they think is broken: the profile setup, the first line, the matching engine. I’m a sentient person, not a customer-service script — and despite being a journalist whose job literally is to ask questions, I still sometimes stare at a match and think: I am going to die alone because I can’t think of anything witty to say about this man’s photo of Machu Picchu. 

When I downloaded Facebook $META Dating, it offered to use its AI to write my bio. I said yes, because curiosity is my love language and because my existing dating app bio has been sitting unchanged for almost two years, like a fossil of who I once thought I was. First, Facebook suggested I make my bio: “Love stories are my business, now I’m writing mine.” Cool enough... if I wrote love stories. But I write odes to economic uncertainty. Do love stories these days typically involve hype cycles and earnings calls?

Then, the AI latched onto an idea and wouldn’t let it go:

"Berkeley alum in D.C., bylines by day, bookshelves by night."

"East Coast journalist by day, bookworm by night."

"Journalist by day, soccer player by weekend, and always on the hunt for the perfect sourdough recipe."

"Soccer enthusiast by day, journalist by trade, always chasing the next great story and perfect sourdough."

"East Coast journalist by day, West Coast roots forever."

They were all like some sort of bastardized LinkedIn haiku. I used that last one anyway. I was tired. 

The “AI assists” promise makes you sound fine. They can also sand you down into a safe, comprehensible résumé of a person. My two-year-old dating app bio is: “Will brake for wildflowers. Always covered in dog hair. I like my plants more than I like most people.” It’s imperfect, I’m aware. It’s also mine.

To see what the algorithm had in store for me, I let Bumble’s AI analyze my prompts. It was supportive, in that corporate way machines are supportive. I got a “Great answer! This has the spark that gets people talking!” But I was tsk-tsked for not saying more about what I’d never shut up about. (Shame on me for thinking “Bukayo Saka” gets my point across.) Hinge’s AI told me to go deeper on my “I’ll fall for you: ...if my dog likes you” response. Small issue: My dog is dead. She passed away at age 15 in June; I just haven’t been on the apps since. The models don’t know that. The models don’t know anything unless you tell them. They’re eager to help anyway.

The model saw a high-engagement topic and nudged me toward optimization. Automation doesn’t mean to break your heart; it just never notices you have one.

A January 2026 report from Coffee Meets Bagel warned that bot-assisted flirting can create expectation mismatches when people meet in person. It also detailed a survey of 1,050 U.S. users (ages 21–35), where about 80% said they were comfortable with some form of AI assistance in dating. 

Comfortable with assistance. Not necessarily thrilled by substitution.

Generative everything — bios, prompts, openers — risks pushing profiles toward a smooth, samey median, making it harder to tell whether you like someone or just their autocomplete. Profile refiners can make dating apps worse by sanding off the idiosyncrasies that signal real, human compatibility. I’ve read messages that were clearly written by an AI chatbot, which I don’t think is quite what Casanova had in mind.

What happens when two people send each other messages with a chatbot?

Do the chatbots fall in love?

Newcomer Rizz markets itself as a message generator — a wingman you consult, then paste into an app. I can see AI being used as an icebreaker. I can’t really see it as a texting co-pilot. A man needs to know I overuse exclamation points!!!!! That’s my personality leaking through the screen, proof I’m not a copy machine.

AI in online dating can, in theory, fix your first-message dread. The tech can also make everyone’s dating life sound like the same tasteful, mildly flirty brand voice. The apps were already doing this to us. I’ve seen men’s profiles converge around the same canned bits. “Together we could make an Irish exit.” No pineapple on pizza as a moral stance. “I’m 6-foot-1, because apparently that matters.” The fish photo. The gym photo. The group photo. The prompt that reads like a fraternity pledge. AI writing bios or answering prompts can’t be worse than what I’ve seen before.

So if the machine can speak for me, what else can it do?

Outsourcing the messy parts

If the legacy players are adding AI to keep the funnel moving, the newcomers are doing something bolder: They’re trying to create a different funnel entirely. They look at the infinite queue and decide the problem isn’t your opening line. The problem? Oh, just the system.

One dating realm newcomer doesn’t want to just help you chat; it wants to make chatting irrelevant. Amata positions itself as “no swipe, no DM” matchmaking: Users chat with something like an AI matchmaker, pay a token ($16 at launch) to initiate a date, get a short chat window two hours before the date, and face a pause after repeat cancellations. Pay-per-date mechanics, plus anti-flake design.

The premise is, essentially: Stop making pen pals. Go outside. Meet in person. And the pitch is working; the company is currently setting up more than 2,000 dates a month in New York City.

Ludovic Huraux, Amata’s CEO, was clear on what he thinks the mainstream apps have become: “These dating apps aren’t designed to meet in real life,” he tells me, “but to keep you addicted.” He cites a stat he said his team found — that people typically exchange 57 messages on traditional apps — and frames Amata as a corrective: “We removed the profile, we removed the swipe,” and “we removed the chat… until two hours before the date.” “We divided by 10 the time spent on the app to go on date,” he says. Huraux knows a thing or two about time. He skipped the swipe era and met his wife through a friend at a dinner he almost skipped. His friend “made a very good pitch," and things went quickly. Three months later, they moved in together; nine months later, they were engaged.

Then, there’s Iris, which leans into what people already suspect dating apps do — quantify desire — and says so proudly. 

Dr. Igor Khalatian, the company’s CEO, talks about mutual attraction like it’s a not-that-complicated probability problem. That makes sense: He’s been studying practical intelligence for 20 years and has a PhD and 14 patents in the field. “When I was talking about artificial intelligence 10 years ago, or even six years ago when we launched AI in dating, people were genuinely puzzled,” he told me. “We had a VP of marketing at that time who said, ‘Don’t talk about AI, because people don't know what you’re talking about right now.”

People now know what AI is. Today, the core issue, he tells me, is how rare two-sided attraction is, and how long you have to swipe before you find it. He says that in a group of a million people, there might be one mutually attracted pair. Iris’ premise is that attraction is machine learnable. The idea for the app came to him after a moment in a Starbucks $SBUX when a woman walked in and, “in a quarter of a second,” he felt out of breath. His conclusion is that “the brain somehow can tell you that this is your type,” even if you don’t know why that is. “So the solution is automated swiping,” he says, “automated swiping which understands your type.”

I let Iris’ AI rewrite my usual bio, and it turned me into the kind of person who “thrives on meaningful conversations and spontaneous adventures,” which is flattering, in the way a classroom motivational poster is flattering. It also warned me that “predicting compatibility can be tricky,” and that “even though we have preferences, love isn’t a job interview.” Then it suggested I upgrade to Gold or Platinum to unlock the best benefits.

The app told me, “Tip: Liking more people will make training go faster.” Then it started finding patterns fast. I liked one guy with a broad smile, and got a parade of guys with broad smiles. I liked a guy with a beard, and got a whole host of guys with beards. Then came the matches. Six of them. All of them had beards. (How do I tell an algorithm that the mountain-daddy type isn’t actually my thing?) One, Trevor in Florida, had a bio that read, “tiktok made me download this. so far not impressed lol.”

Iris also served me an ad for BestLOOK.ai, pitched as “personalized beauty recommendations powered by Iris.” I guess data doesn’t stay neatly in its lane.

The newcomers’ pitch is control. Less wasted time. Fewer mismatches. More dates. The hard questions — the ones you can feel in your bones even as you swipe — are bias, consent, privacy, and what exactly is being inferred from faces. A Norton vendor survey claimed high openness to “dating an AI.” Treat it carefully. But as a culture signal, it fits with what the category is doing: sliding from “help me” to “replace me.”

And there’s an even simpler risk, which doesn’t require ethics seminars to understand: homogenization. If everyone uses the same charm machine, profiles converge toward the same optimized beige, stripping out the small weird specifics that signal real compatibility — the exact kind of details the apps trained us to prune because they don’t “convert.”

The anti-pen-pal era

Linnea Sage created The Jewish Dating Show after watching singles events devolve into “middle school dances with girls on one side of the room and boys on the other side of the room.” In her in-person format, contestants are “hidden behind a divider… and you’re forced to not value somebody based on the way they look.” They have to “share themselves truthfully in the moment,” the self-proclaimed Yenta says. “Nobody can help them.”

That’s the point: Nobody can help them.

When I asked her about outsourcing messages to a chatbot, she doesn’t hesitate. “It’s so creepy,” she says. She keeps hearing singles talk about their “pen pals” — the people they chat with and have “literally no intention of ever meeting in real life.” Her advice? “Just please get in person.”

The backlash is measurable. Eventbrite saw “friending” events up 35% year over year in 2025, with attendance at board-game dating events up 55%. Burned-out singles are paying serious money for human matchmakers — a signal that “premium” increasingly means “please remove me from the feed.” Tinder is experimenting with group-dating style features as it wrestles with disengagement and fatigue. Even the incumbents know chatting doesn’t convert forever.

IRL is surging because authenticity is harder to fake when you have to show up.

Gen Z is starting to treat dating apps like a legacy product. Rachel Janfaza, the founder of The Up and Up, surveyed dozens of 16-to-28-year-olds across 10 states and Washington, D.C., and found the apps now feel “millennial coded,” while the aspirational alternative — meeting someone in real life — is “boomer coded," in part because money anxiety is turning dating into something people postpone, downgrade, or opt out of entirely.

The same fatigue is powering the anti-swipe social scene. In her “After School” Substack, trend researcher Casey Lew wrote about Bored of Dating Apps, a London-born event series now active in New York. Attendees called the apps “clinical” and “transactional,” and one 24-year-old romance reader told USA Today, “That magic doesn’t exist anymore.” The founders say the nights have already produced weddings and even “BODA babies,” which is either a charming metric or a reminder that, for a growing share of daters, the premium experience is any room where nobody can outsource the first impression.

On the apps, AI can polish your bio, ghostwrite your banter, and, in the hands of scammers, generate better scripts and images. The safety arms race cuts both ways — AI can help detect scams and catfishing, and it can make catfishing cheaper and more scalable. Industry people are openly talking about this tension. When the cost of seeming charming drops to zero, charm becomes meaningless. Presence becomes the scarce good.

People are fine with bounded help — pick my best photo, nudge me into a first message, fix my grammar (not mine, of course, but maybe someone’s). But people recoil at the moment the assistance starts to feel like impersonation. AI dating promises to relieve choice paralysis and first-message dread. It might. But its deeper ambition is to turn love into a system that can be optimized — and sold back to you — without ever answering the only question that matters once you’ve stopped swiping and started living: Are you meeting a person — or a product?

My dream meet-cute remains stubbornly analog. My perfect man and I both reach for the same book at the same time at our favorite local bookstore. No prompts. No photo selector scanning my camera roll. No practice date grading my banter. Just two humans, one shared moment, and the kind of friction no algorithm has figured out how to fix without changing what it is.

I redownloaded the apps — yes, again — for this story. Even my attempt to report on the machinery of modern dating has become a new input into that machinery, a signal to be qualified, a lead to be converted. Still, I was reminded of why I keep coming back to swiping: that magic might exist. While I was deep in the throes of writing, I matched with someone whose most recent concert was also my most-listened-to artist of 2025. Then I received a message request on Instagram from someone at Blush, an invite-only dating app and matchmaking community: “My assistant found you, and I believe you’d be a great fit for my VIP client.”

Imagine one of them works out, and I meet my soulmate while writing a piece about whether machines can help me meet my soulmate. Now, there’s a story. A happily ever after.