Love in the Age of AI

We were convinced we had already reached a sufficiently advanced stage in the transformation of love into a digital problem. Opening Tinder, Hinge or Bumble, swiping through photos of people just as quickly as we compare hotels on Booking, deciding in half a second whether or not someone deserves to enter our lives, entrusting an algorithm with the task of showing us who we might like: that already seemed like enough. Evidently, we were only just getting started.

Justin McLeod, the founder of Hinge, has left the company he created to set up Overtone, an AI-powered matchmaking service. He has raised $18 million, with support from Match Group – which owns, amongst others, Tinder and Hinge itself – and has brought the renowned psychotherapist Esther Perel onto the board of directors. Overtone is expected to launch initially in just a few cities by the end of the year. And its promise is simple: no more wasting time choosing. McLeod even insists that Overtone “is not a dating app”. And indeed, the idea is almost the opposite of traditional apps. No endless catalogue of photographs, no dozens of simultaneous matches, no compulsive swiping. The AI converses with the user, listens to their voice, pieces together their story, and tries to understand not simply how old they are or what they do for a living, but what sort of person they are, what they want, what relationships they’ve had, and what they’re looking for in their next one. It then selects a few people deemed compatible and even explains why it thinks they’re worth meeting. The chemistry, McLeod magnanimously concedes, will still need to be tested in person. For now, at least. The idea is fascinating because it uses AI to solve a problem caused in part by previous technology: an overload of choice. The old apps gave us the impression that behind every person there was always a better one, just another swipe away. The new generation wants to tell us exactly the opposite: stop searching, we’ll take care of it. But to understand just how far we can go, we need to look to China. Liangpei, which roughly translates as ‘good match’, was launched in July and offers a deal that would put even marriage agencies to shame: pay 2,000 yuan, around 275 dollars, and if you don’t get married within three years, they’ll give you your money back. This guarantee is particularly remarkable in a country where 6.76 million couples got married in 2025 – roughly half the number compared with ten years earlier.

Here, too, the process begins with a conversation with a machine. Users spend twenty minutes talking to an artificial matchmaker, answering twenty-five questions about their romantic history, marriage plans, and even intentions to buy a home. The algorithm builds a profile, assigns compatibility scores and presents potential matches. But Liangpei introduces an even more revolutionary rule: once two people are matched, they can only chat with each other until one of them decides to end the conversation. No parallel market, no five chats at once, no temptation to check whether someone with a better score has come along in the meantime. “We want to get users off the platform as soon as possible,” says founder Zeng Xinxun. An almost subversive business model: an app that dreams of losing its customers. Liangpei already has more than 9,300 active users across 300 cities. Around one in ten has secured an exclusive match; conversations last on average more than ten days and exceed 200 messages. The AI also suggests questions when the conversation stalls and can act as a ‘wingman’, analysing the exchange and offering advice, although for now it doesn’t write the replies directly. And in Shenzhen, it can rank, out of 1,500 users, the ten who are most likely to be a good match for you. The fascinating frontier of romantic AI is therefore not the chatbot you fall in love with. It is something more subtle: the chatbot that decides who we should fall in love with. We may find this unsettling, but we must recognise that the question it attempts to answer is very human. We have always asked friends, relatives and matchmakers: ‘Who do you think would be right for me?’. Now we ask a machine that has thousands of candidates, twenty-five responses, a detailed biography and a statistical model. One small detail remains, one on which almost all Western literature has been based for a few centuries: people have an annoying tendency to fall in love with the wrong person. And it is possible that the final major hurdle for AI will not be building an algorithm capable of finding the perfect partner. It will be convincing two human beings that the perfect partner really is the one the algorithm has chosen for them.