PARIS — In 2017, when a journalist asked the co-founder of Tinder how he imagined his dating app in five years' time, Sean Rad pulled out his smartphone and pretended to have a conversation with the device.

"The Tinder voice might pop up and say, ‘There's someone down the street that we think you're going to be attracted to, and she's also attracted to you, and guess what, she's free tomorrow night! And we know you both like this indie band, and it's playing, so would you like us to buy you tickets?'"

The prospect of algorithms finding your perfect match by compiling enormous amounts of data — from age, gender, location and sexual preferences, to online purchasing history and even Spotify playlists — remains a fantasy. But artificial intelligence (AI) is progressively entering the field of dating.

Back in 2014, the Canadian computer programmer Justin Long started working on, an algorithm which deduced users' physical preferences, based on their swiping history on Tinder. The robot, which is no longer available, used facial recognition to select profiles on behalf of the user. If there was a match, it started the conversation with an opening line. Long claims that hundreds of thousands of people received the opener, "Do you like avocados?"

The start-up Loveflutter relies on Twitter and Facebook posts to match users. By teaming up with, a company that specializes in the automated analysis of people's psychological states based on language use, the dating app promises to connect compatible personalities.

With machines that track our behavior replacing questionnaires about our preferences, dating services are changing the game.

A market based on illusion.

The scenarios imagined by the television series Black Mirror, or by Silicon Valley's Peter Diamandis — who believes that virtual personal assistants will be able to find your "perfect partner" in the very near future — seem closer than ever. But they are not to everybody's liking.

"It's an enormous market that's being created, but it is based upon an illusion," says French sociologist Jean-Claude Kaufmann. "Algorithms are completely incapable of finding our ideal partner, as is our Internet history. We are victims of a collective belief in this domain."

Another sociologist Catherine Lejealle doubts that "the Netflix formula" can be adapted for relationships. "It sells something that can't be put into an algorithm; as we get older, we look for different things. Things such as our pace of life, which are an essential part of getting along with somebody on a daily basis, are not considered."

In a society based upon rational control, technology can be extremely useful for making decisions. Yet romantic relationships are far more complex. "Our objective is not to solve questions of the heart using a large matrix filled with millions of pieces of data on each individual and working with algorithms," insists Xavier de Baillenx, who heads innovation at French dating site Meetic.

German man using Tinder, his collection of matches referred to as "Tindergarten" — Photo: Paul Zinken/DPA/ZUMA

Intention detection

De Baillenx developed the personal assistant Lara while co-running the start-up Pretty Fun Therapy, which has since been bought by Meetic, and he is now committed to improving it while working for one of Europe's largest dating sites. "What we do is 'intention detection,' not necessarily ‘matching'. We are there to save people time and to give them the tools to meet people."

Today, Lara has around 50 distinct pieces of advice to give, suggesting locations for a date, what to wear, or how to overcome shyness. The personal assistant also accompanies the users as they create their account, and the team is working to improve her understanding and memory to allow her to hold a conversation.

At Happn, where AI is being gradually integrated into the matching algorithm, "the first development lead to double the amount of ‘crushes'" — the Happn equivalent of a match. Thanks to the two billion pieces of data that are processed by the app every day (GPS coordinates, information given by the user or taken from Facebook), "series of correlations that are beyond even our understanding" help to select the most relevant profiles, according to CEO Didier Rappaport.

The notion of chance could be removed from dating.

Such a process comes with its fair share of risks. "Artificial Intelligence is not neutral," says Marina Pavlovic Rivas, an independent advanced analytics consultant. "It feeds off the information we give it, which can include biases that the algorithm will then generalize. In a society ruled by big data and algorithms, AI could, for example, deduce socio-economic status from a ZIP code and, identifying a degree of endogamy within its data, propose only profiles coming from the same background as its users. The same could happen for skin color.

The notion of chance could thus be removed from dating, by proposing only those who are a match according to an algorithm susceptible to automate certain forms of discrimination. "We have a responsibility to avoid the algorithm's determinism," says Guillaume Perrin-Houdon, Chief Data Officer at Happn. Despite AI's utility, it is the human's job to "make the machine moral", according to Pavlovic Rivas.

The initiatives put in place include forcing the AI to propose profiles that do not have the best results, and looking for previously unexplored alternatives, such as proposing a redhead if the user has only been seeing blonds. Because for the moment, the heart still reasons in ways that artificial intelligence may never understand.

See more from Tech / Science here