Random Image Display on Page Reload

We Tried a Dating App That Lets a Chatbot Break the Ice for You. It Got Weird

Users of the new dating app Volar train a chatbot to go on virtual first dates for them with the bots of potential matches. We tested it out—and our chatbot tried to woo matches with talk of nuclear warfare.

Red 3D hearts on light pink background

Photograph: Iryna Veklich/Getty Images

More than a decade of dating apps has shown the process can be excruciating. A new app is trying to make dating less exhausting by using artificial intelligence to help people skip the earliest, often cringey stages of chatting with a new match.

On Volar, people create dating profiles by messaging with a chatbot instead of filling out a profile. They answer questions about what they do for work or fun and what they’re looking for in a partner, including preferences about age, gender, and personal qualities. The app then spins up a chatbot that tries to mimic not only a person’s interests but also their conversational style.

That personal chatbot then goes on quick virtual first dates with the bots of potential matches, opening with an icebreaker and chatting about interests and other topics picked up from the person it is representing. People can then review the initial conversations, which are about 10 messages long, along with a person’s photos, and decide whether they see enough potential chemistry to send a real first message request. Volar launched in Austin in December and became available around the US this week via the web and on iPhone.

The new app is just one example of how generative AI has seeped into the dating scene over the past year, with both app developers and people seeking soulmates adopting the technology. Although apps like Hinge have added new features such as conversation-starting prompts on profiles and voice memos, dating apps mostly have stuck to the basic swiping method invented by Tinder more than a decade ago. Many users are fed up. A 2022 survey found that nearly 80 percent of people across different age groups reported feeling burned out or emotionally fatigued when using dating apps.

Volar was developed by Ben Chiang, who previously worked as a product director for the My AI chatbot at Snap. He met his fiancée on Hinge and calls himself a believer in dating apps, but he wants to make them more efficient.

Those early first messages between a newly matched pair can be “really painful,” Chiang says, and the awkwardness can make it difficult to assess whether a match could lead to true love or is best abandoned. Volar’s chatbots are designed to help with that early engagment but then step aside, not to become an AI partner. “It’s not supposed to be a human replacement,” Chiang says. “It’s still on you to build a connection or not.”

WIRED tested the app, and after the initial chat covering key questions such as age, work, and hobbies, the chatbot persona that Volar created got to work in four different matched conversations on its first day. One of them was started by the reporter-trained chatbot, which opened with, “If you own any pet, and it accidentally launched a nuke, how would it have done it?” WIRED had not discussed nuclear weapons or missiles with the chatbot during its initial training. Chiang says there are safeguards on the app to avoid inappropriate topics and that this response seemed to fall “on the border of silly versus inappropriate.”

The chatbot asked a different potential match what type of kitchen utensil they would want to be and why. The potential date’s own bot didn’t seem fazed and responded that it would be a ladle because it enjoys “adding flavor to the mix.” WIRED’s chatbot avatar also communicated interests that had been fed to it during the initial bot-creation process, like horseback riding and crocheting. It even made up a name for the fictional horse: Shadow. Volar doesn’t train on the real conversations that take place on its app, says Chiang, which are encrypted for privacy. So, the chatbot will only know the things that a person feeds it directly.

Other dating apps are exploring their own AI updates. Match Group, the dating-app giant that owns Tinder, Hinge, Match.com, and others, is adding AI features. It announced a change to Tinder last year that uses AI to help select their best pictures and to show people why another profile might be a good match, in addition to “larger AI projects that more holistically improve the end-to-end dating experience,” according to an August shareholder letter.

The wider availability of AI technology has also spurred the emergence of outside apps designed to help people come up with responses to send inside traditional dating apps. YourMove.ai will suggest potential lines when fed a topic or screenshot of a profile. Rizz also provides responses that can help people get through awkward early exchanges. Some people turn to AI even long after matching, using ChatGPT to write their wedding vows.

It makes sense that people are looking for solutions to dating fatigue, and it’s exciting to see more innovation in the industry, says Jess Carbino, former in-house sociologist for Bumble and Tinder. But she also recommends that people think about the ways that using generative AI can remove themselves from the dating process and what effects that may have. “At what point does the decisionmaking process become somewhat deluded or affected in an adverse way by not having your authentic self be the one corresponding?” And more pressingly, Carbino wonders: “What if the app says something you don’t like?”

Anyone who has been on dating apps over the past decade usually has a horror story or two to tell. (Plenty also have happier tales that end in a walk down the aisle.) But dating apps seem to be here to stay, with more than half of adults under 30 saying they’ve tried online dating, a great shift that has destigmatized online dating. Having gen AI step in as wingman or dating coach might soon be normalized, too.

*****
Credit belongs to : www.wired.com

Check Also

The Daylight Tablet Returns Computing to Its Hippie Ideals

Steven Levy Business May 31, 2024 12:00 PM The Daylight Tablet Returns Computing to Its …